detection.py 75.7 KB
Newer Older
1
#  Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserve.
2 3 4 5 6
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
7
#    http://www.apache.org/licenses/LICENSE-2.0
8 9 10 11 12 13 14 15 16 17
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
All layers just related to the detection neural network.
"""

18 19
from __future__ import print_function

20 21
from .layer_function_generator import generate_layer_fn
from .layer_function_generator import autodoc, templatedoc
22
from ..layer_helper import LayerHelper
D
dengkaipeng 已提交
23
from ..framework import Variable
24 25
from . import tensor
from . import nn
26
from . import ops
M
minqiyang 已提交
27
from ... import compat as cpt
C
chengduoZH 已提交
28
import math
M
minqiyang 已提交
29
import six
30
import numpy
31
from functools import reduce
32

C
chengduoZH 已提交
33
__all__ = [
J
jerrywgz 已提交
34 35 36 37 38
    'prior_box', 'density_prior_box', 'multi_box_head', 'bipartite_match',
    'target_assign', 'detection_output', 'ssd_loss', 'detection_map',
    'rpn_target_assign', 'anchor_generator', 'roi_perspective_transform',
    'generate_proposal_labels', 'generate_proposals', 'iou_similarity',
    'box_coder', 'polygon_box_transform', 'yolov3_loss', 'box_clip'
C
chengduoZH 已提交
39
]
40 41


42 43
def rpn_target_assign(bbox_pred,
                      cls_logits,
Y
Yuan Gao 已提交
44
                      anchor_box,
45
                      anchor_var,
46 47 48
                      gt_boxes,
                      is_crowd,
                      im_info,
Y
Yuan Gao 已提交
49
                      rpn_batch_size_per_im=256,
50 51
                      rpn_straddle_thresh=0.0,
                      rpn_fg_fraction=0.5,
Y
Yuan Gao 已提交
52
                      rpn_positive_overlap=0.7,
53 54
                      rpn_negative_overlap=0.3,
                      use_random=True):
Y
Yuan Gao 已提交
55
    """
H
haowang101779990 已提交
56
    **Target Assign Layer for region proposal network (RPN) in Faster-RCNN detection.**
Y
Yuan Gao 已提交
57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

    This layer can be, for given the  Intersection-over-Union (IoU) overlap
    between anchors and ground truth boxes, to assign classification and
    regression targets to each each anchor, these target labels are used for
    train RPN. The classification targets is a binary class label (of being
    an object or not). Following the paper of Faster-RCNN, the positive labels
    are two kinds of anchors: (i) the anchor/anchors with the highest IoU
    overlap with a ground-truth box, or (ii) an anchor that has an IoU overlap
    higher than rpn_positive_overlap(0.7) with any ground-truth box. Note
    that a single ground-truth box may assign positive labels to multiple
    anchors. A non-positive anchor is when its IoU ratio is lower than
    rpn_negative_overlap (0.3) for all ground-truth boxes. Anchors that are
    neither positive nor negative do not contribute to the training objective.
    The regression targets are the encoded ground-truth boxes associated with
    the positive anchors.

    Args:
74
        bbox_pred(Variable): A 3-D Tensor with shape [N, M, 4] represents the
Y
Yuan Gao 已提交
75 76 77
            predicted locations of M bounding bboxes. N is the batch size,
            and each bounding box has four coordinate values and the layout
            is [xmin, ymin, xmax, ymax].
78 79 80
        cls_logits(Variable): A 3-D Tensor with shape [N, M, 1] represents the
            predicted confidence predictions. N is the batch size, 1 is the
            frontground and background sigmoid, M is number of bounding boxes.
Y
Yuan Gao 已提交
81 82 83 84 85 86
        anchor_box(Variable): A 2-D Tensor with shape [M, 4] holds M boxes,
            each box is represented as [xmin, ymin, xmax, ymax],
            [xmin, ymin] is the left top coordinate of the anchor box,
            if the input is image feature map, they are close to the origin
            of the coordinate system. [xmax, ymax] is the right bottom
            coordinate of the anchor box.
87 88
        anchor_var(Variable): A 2-D Tensor with shape [M,4] holds expanded 
            variances of anchors.
89
        gt_boxes (Variable): The ground-truth boudding boxes (bboxes) are a 2D
Y
Yuan Gao 已提交
90 91
            LoDTensor with shape [Ng, 4], Ng is the total number of ground-truth
            bboxes of mini-batch input.
92 93 94
        is_crowd (Variable): A 1-D LoDTensor which indicates groud-truth is crowd.
        im_info (Variable): A 2-D LoDTensor with shape [N, 3]. N is the batch size,
        3 is the height, width and scale.
Y
Yuan Gao 已提交
95
        rpn_batch_size_per_im(int): Total number of RPN examples per image.
96 97 98
        rpn_straddle_thresh(float): Remove RPN anchors that go outside the image
            by straddle_thresh pixels.
        rpn_fg_fraction(float): Target fraction of RoI minibatch that is labeled
Y
Yuan Gao 已提交
99 100 101 102 103 104 105 106 107
            foreground (i.e. class > 0), 0-th class is background.
        rpn_positive_overlap(float): Minimum overlap required between an anchor
            and ground-truth box for the (anchor, gt box) pair to be a positive
            example.
        rpn_negative_overlap(float): Maximum overlap allowed between an anchor
            and ground-truth box for the (anchor, gt box) pair to be a negative
            examples.

    Returns:
M
minqiyang 已提交
108
        tuple:
Y
Yuan Gao 已提交
109
               A tuple(predicted_scores, predicted_location, target_label,
J
jerrywgz 已提交
110 111
               target_bbox, bbox_inside_weight) is returned. The predicted_scores 
               and predicted_location is the predicted result of the RPN.
Y
Yuan Gao 已提交
112 113 114 115 116 117 118
               The target_label and target_bbox is the ground truth,
               respectively. The predicted_location is a 2D Tensor with shape
               [F, 4], and the shape of target_bbox is same as the shape of
               the predicted_location, F is the number of the foreground
               anchors. The predicted_scores is a 2D Tensor with shape
               [F + B, 1], and the shape of target_label is same as the shape
               of the predicted_scores, B is the number of the background
M
minqiyang 已提交
119
               anchors, the F and B is depends on the input of this operator.
J
jerrywgz 已提交
120 121
               Bbox_inside_weight represents whether the predicted loc is fake_fg
               or not and the shape is [F, 4].
Y
Yuan Gao 已提交
122 123 124 125

    Examples:
        .. code-block:: python

H
haowang101779990 已提交
126 127 128 129 130 131 132 133 134 135 136 137 138 139
            bbox_pred = layers.data(name='bbox_pred', shape=[100, 4],
                              append_batch_size=False, dtype='float32')
            cls_logits = layers.data(name='cls_logits', shape=[100, 1],
                              append_batch_size=False, dtype='float32')
            anchor_box = layers.data(name='anchor_box', shape=[20, 4],
                              append_batch_size=False, dtype='float32')
            gt_boxes = layers.data(name='gt_boxes', shape=[10, 4],
                             append_batch_size=False, dtype='float32')
            loc_pred, score_pred, loc_target, score_target, bbox_inside_weight =
                fluid.layers.rpn_target_assign(bbox_pred=bbox_pred,
                                              cls_logits=cls_logits,
                                              anchor_box=anchor_box,
                                              gt_boxes=gt_boxes)

Y
Yuan Gao 已提交
140 141 142
    """

    helper = LayerHelper('rpn_target_assign', **locals())
143
    # Assign target label to anchors
J
jerrywgz 已提交
144 145 146 147 148 149 150
    loc_index = helper.create_variable_for_type_inference(dtype='int32')
    score_index = helper.create_variable_for_type_inference(dtype='int32')
    target_label = helper.create_variable_for_type_inference(dtype='int32')
    target_bbox = helper.create_variable_for_type_inference(
        dtype=anchor_box.dtype)
    bbox_inside_weight = helper.create_variable_for_type_inference(
        dtype=anchor_box.dtype)
Y
Yuan Gao 已提交
151 152
    helper.append_op(
        type="rpn_target_assign",
153 154 155 156 157 158
        inputs={
            'Anchor': anchor_box,
            'GtBoxes': gt_boxes,
            'IsCrowd': is_crowd,
            'ImInfo': im_info
        },
Y
Yuan Gao 已提交
159 160 161
        outputs={
            'LocationIndex': loc_index,
            'ScoreIndex': score_index,
162
            'TargetLabel': target_label,
J
jerrywgz 已提交
163
            'TargetBBox': target_bbox,
J
jerrywgz 已提交
164
            'BBoxInsideWeight': bbox_inside_weight
Y
Yuan Gao 已提交
165 166 167
        },
        attrs={
            'rpn_batch_size_per_im': rpn_batch_size_per_im,
168
            'rpn_straddle_thresh': rpn_straddle_thresh,
Y
Yuan Gao 已提交
169 170
            'rpn_positive_overlap': rpn_positive_overlap,
            'rpn_negative_overlap': rpn_negative_overlap,
171 172
            'rpn_fg_fraction': rpn_fg_fraction,
            'use_random': use_random
Y
Yuan Gao 已提交
173 174
        })

175 176 177 178
    loc_index.stop_gradient = True
    score_index.stop_gradient = True
    target_label.stop_gradient = True
    target_bbox.stop_gradient = True
J
jerrywgz 已提交
179
    bbox_inside_weight.stop_gradient = True
Y
Yuan Gao 已提交
180

181 182 183 184
    cls_logits = nn.reshape(x=cls_logits, shape=(-1, 1))
    bbox_pred = nn.reshape(x=bbox_pred, shape=(-1, 4))
    predicted_cls_logits = nn.gather(cls_logits, score_index)
    predicted_bbox_pred = nn.gather(bbox_pred, loc_index)
185

J
jerrywgz 已提交
186
    return predicted_cls_logits, predicted_bbox_pred, target_label, target_bbox, bbox_inside_weight
Y
Yuan Gao 已提交
187 188


Y
Yuan Gao 已提交
189 190
def detection_output(loc,
                     scores,
191 192 193 194 195 196 197 198 199
                     prior_box,
                     prior_box_var,
                     background_label=0,
                     nms_threshold=0.3,
                     nms_top_k=400,
                     keep_top_k=200,
                     score_threshold=0.01,
                     nms_eta=1.0):
    """
200
    **Detection Output Layer for Single Shot Multibox Detector (SSD).**
201

202 203
    This operation is to get the detection results by performing following
    two steps:
C
caoying03 已提交
204

205 206 207 208 209 210
    1. Decode input bounding box predictions according to the prior boxes.
    2. Get the final detection results by applying multi-class non maximum
       suppression (NMS).

    Please note, this operation doesn't clip the final output bounding boxes
    to the image window.
211 212 213 214 215 216

    Args:
        loc(Variable): A 3-D Tensor with shape [N, M, 4] represents the
            predicted locations of M bounding bboxes. N is the batch size,
            and each bounding box has four coordinate values and the layout
            is [xmin, ymin, xmax, ymax].
Y
Yuan Gao 已提交
217 218 219 220
        scores(Variable): A 3-D Tensor with shape [N, M, C] represents the
            predicted confidence predictions. N is the batch size, C is the
            class number, M is number of bounding boxes. For each category
            there are total M scores which corresponding M bounding boxes.
221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
        prior_box(Variable): A 2-D Tensor with shape [M, 4] holds M boxes,
            each box is represented as [xmin, ymin, xmax, ymax],
            [xmin, ymin] is the left top coordinate of the anchor box,
            if the input is image feature map, they are close to the origin
            of the coordinate system. [xmax, ymax] is the right bottom
            coordinate of the anchor box.
        prior_box_var(Variable): A 2-D Tensor with shape [M, 4] holds M group
            of variance.
        background_label(float): The index of background label,
            the background label will be ignored. If set to -1, then all
            categories will be considered.
        nms_threshold(float): The threshold to be used in NMS.
        nms_top_k(int): Maximum number of detections to be kept according
            to the confidences aftern the filtering detections based on
            score_threshold.
        keep_top_k(int): Number of total bboxes to be kept per image after
            NMS step. -1 means keeping all bboxes after NMS step.
        score_threshold(float): Threshold to filter out bounding boxes with
            low confidence score. If not provided, consider all boxes.
        nms_eta(float): The parameter for adaptive NMS.

    Returns:
M
minqiyang 已提交
243 244
        Variable:

245
            The detection outputs is a LoDTensor with shape [No, 6].
246 247 248 249 250 251 252 253
            Each row has six values: [label, confidence, xmin, ymin, xmax, ymax].
            `No` is the total number of detections in this mini-batch. For each
            instance, the offsets in first dimension are called LoD, the offset
            number is N + 1, N is the batch size. The i-th image has
            `LoD[i + 1] - LoD[i]` detected results, if it is 0, the i-th image
            has no detected results. If all images have not detected results,
            all the elements in LoD are 0, and output tensor only contains one
            value, which is -1.
254 255 256 257

    Examples:
        .. code-block:: python

258
            pb = layers.data(name='prior_box', shape=[10, 4],
259
                         append_batch_size=False, dtype='float32')
260
            pbv = layers.data(name='prior_box_var', shape=[10, 4],
261
                          append_batch_size=False, dtype='float32')
262
            loc = layers.data(name='target_box', shape=[2, 21, 4],
263
                          append_batch_size=False, dtype='float32')
264
            scores = layers.data(name='scores', shape=[2, 21, 10],
265
                          append_batch_size=False, dtype='float32')
266
            nmsed_outs = fluid.layers.detection_output(scores=scores,
267 268 269 270 271
                                       loc=loc,
                                       prior_box=pb,
                                       prior_box_var=pbv)
    """
    helper = LayerHelper("detection_output", **locals())
272 273 274 275 276
    decoded_box = box_coder(
        prior_box=prior_box,
        prior_box_var=prior_box_var,
        target_box=loc,
        code_type='decode_center_size')
277
    scores = nn.softmax(input=scores)
Y
Yuan Gao 已提交
278
    scores = nn.transpose(scores, perm=[0, 2, 1])
279
    scores.stop_gradient = True
X
Xin Pan 已提交
280 281
    nmsed_outs = helper.create_variable_for_type_inference(
        dtype=decoded_box.dtype)
282 283 284 285 286 287 288 289 290 291 292 293 294
    helper.append_op(
        type="multiclass_nms",
        inputs={'Scores': scores,
                'BBoxes': decoded_box},
        outputs={'Out': nmsed_outs},
        attrs={
            'background_label': 0,
            'nms_threshold': nms_threshold,
            'nms_top_k': nms_top_k,
            'keep_top_k': keep_top_k,
            'score_threshold': score_threshold,
            'nms_eta': 1.0
        })
295
    nmsed_outs.stop_gradient = True
296
    return nmsed_outs
C
chengduoZH 已提交
297 298


X
Xin Pan 已提交
299 300 301 302 303 304 305 306 307 308 309 310 311 312
@templatedoc()
def iou_similarity(x, y, name=None):
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        y(${y_type}): ${y_comment}

    Returns:
        out(${out_type}): ${out_comment}
    """
    helper = LayerHelper("iou_similarity", **locals())
    if name is None:
X
Xin Pan 已提交
313
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
X
Xin Pan 已提交
314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="iou_similarity",
        inputs={"X": x,
                "Y": y},
        attrs={},
        outputs={"Out": out})
    return out


@templatedoc()
def box_coder(prior_box,
              prior_box_var,
              target_box,
              code_type="encode_center_size",
              box_normalized=True,
              name=None):
    """
    ${comment}

    Args:
        prior_box(${prior_box_type}): ${prior_box_comment}
        prior_box_var(${prior_box_var_type}): ${prior_box_var_comment}
        target_box(${target_box_type}): ${target_box_comment}
        code_type(${code_type_type}): ${code_type_comment}
        box_normalized(${box_normalized_type}): ${box_normalized_comment}

    Returns:
        output_box(${output_box_type}): ${output_box_comment}
    """
    helper = LayerHelper("box_coder", **locals())

    if name is None:
X
Xin Pan 已提交
350 351
        output_box = helper.create_variable_for_type_inference(
            dtype=prior_box.dtype)
X
Xin Pan 已提交
352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381
    else:
        output_box = helper.create_variable(
            name=name, dtype=prior_box.dtype, persistable=False)

    helper.append_op(
        type="box_coder",
        inputs={
            "PriorBox": prior_box,
            "PriorBoxVar": prior_box_var,
            "TargetBox": target_box
        },
        attrs={"code_type": code_type,
               "box_normalized": box_normalized},
        outputs={"OutputBox": output_box})
    return output_box


@templatedoc()
def polygon_box_transform(input, name=None):
    """
    ${comment}

    Args:
        input(${input_type}): ${input_comment}

    Returns:
        output(${output_type}): ${output_comment}
    """
    helper = LayerHelper("polygon_box_transform", **locals())
    if name is None:
X
Xin Pan 已提交
382
        output = helper.create_variable_for_type_inference(dtype=input.dtype)
X
Xin Pan 已提交
383 384 385 386 387 388 389 390 391 392 393 394
    else:
        output = helper.create_variable(
            name=name, dtype=prior_box.input, persistable=False)

    helper.append_op(
        type="polygon_box_transform",
        inputs={"Input": input},
        attrs={},
        outputs={"Output": output})
    return output


D
dengkaipeng 已提交
395 396 397
@templatedoc(op_type="yolov3_loss")
def yolov3_loss(x,
                gtbox,
D
dengkaipeng 已提交
398
                gtlabel,
D
dengkaipeng 已提交
399 400 401
                anchors,
                class_num,
                ignore_thresh,
D
dengkaipeng 已提交
402 403 404 405 406
                loss_weight_xy=None,
                loss_weight_wh=None,
                loss_weight_conf_target=None,
                loss_weight_conf_notarget=None,
                loss_weight_class=None,
D
dengkaipeng 已提交
407 408 409 410 411 412
                name=None):
    """
    ${comment}

    Args:
        x (Variable): ${x_comment}
D
dengkaipeng 已提交
413 414 415 416 417 418 419
        gtbox (Variable): groud truth boxes, should be in shape of [N, B, 4],
                          in the third dimenstion, x, y, w, h should be stored 
                          and x, y, w, h should be relative value of input image.
                          N is the batch number and B is the max box number in 
                          an image.
        gtlabel (Variable): class id of ground truth boxes, shoud be ins shape
                            of [N, B].
D
dengkaipeng 已提交
420 421 422
        anchors (list|tuple): ${anchors_comment}
        class_num (int): ${class_num_comment}
        ignore_thresh (float): ${ignore_thresh_comment}
D
dengkaipeng 已提交
423 424 425 426 427
        loss_weight_xy (float|None): ${loss_weight_xy_comment}
        loss_weight_wh (float|None): ${loss_weight_wh_comment}
        loss_weight_conf_target (float|None): ${loss_weight_conf_target_comment}
        loss_weight_conf_notarget (float|None): ${loss_weight_conf_notarget_comment}
        loss_weight_class (float|None): ${loss_weight_class_comment}
D
dengkaipeng 已提交
428 429 430 431 432 433 434 435
        name (string): the name of yolov3 loss

    Returns:
        Variable: A 1-D tensor with shape [1], the value of yolov3 loss

    Raises:
        TypeError: Input x of yolov3_loss must be Variable
        TypeError: Input gtbox of yolov3_loss must be Variable"
D
dengkaipeng 已提交
436
        TypeError: Input gtlabel of yolov3_loss must be Variable"
D
dengkaipeng 已提交
437 438 439 440 441 442 443
        TypeError: Attr anchors of yolov3_loss must be list or tuple
        TypeError: Attr class_num of yolov3_loss must be an integer
        TypeError: Attr ignore_thresh of yolov3_loss must be a float number

    Examples:
    .. code-block:: python

D
dengkaipeng 已提交
444 445 446
        x = fluid.layers.data(name='x', shape=[255, 13, 13], dtype='float32')
        gtbox = fluid.layers.data(name='gtbox', shape=[6, 5], dtype='float32')
        gtlabel = fluid.layers.data(name='gtlabel', shape=[6, 1], dtype='int32')
D
dengkaipeng 已提交
447 448 449 450 451 452 453 454 455 456
        anchors = [10, 13, 16, 30, 33, 23]
        loss = fluid.layers.yolov3_loss(x=x, gtbox=gtbox, class_num=80
                                        anchors=anchors, ignore_thresh=0.5)
    """
    helper = LayerHelper('yolov3_loss', **locals())

    if not isinstance(x, Variable):
        raise TypeError("Input x of yolov3_loss must be Variable")
    if not isinstance(gtbox, Variable):
        raise TypeError("Input gtbox of yolov3_loss must be Variable")
D
dengkaipeng 已提交
457 458
    if not isinstance(gtlabel, Variable):
        raise TypeError("Input gtlabel of yolov3_loss must be Variable")
D
dengkaipeng 已提交
459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
    if not isinstance(anchors, list) and not isinstance(anchors, tuple):
        raise TypeError("Attr anchors of yolov3_loss must be list or tuple")
    if not isinstance(class_num, int):
        raise TypeError("Attr class_num of yolov3_loss must be an integer")
    if not isinstance(ignore_thresh, float):
        raise TypeError(
            "Attr ignore_thresh of yolov3_loss must be a float number")

    if name is None:
        loss = helper.create_variable_for_type_inference(dtype=x.dtype)
    else:
        loss = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    attrs = {
        "anchors": anchors,
        "class_num": class_num,
        "ignore_thresh": ignore_thresh,
    }

D
dengkaipeng 已提交
479 480 481 482 483 484 485 486 487 488 489 490
    if loss_weight_xy is not None and isinstance(loss_weight_xy, float):
        self.attrs['loss_weight_xy'] = loss_weight_xy
    if loss_weight_wh is not None and isinstance(loss_weight_wh, float):
        self.attrs['loss_weight_wh'] = loss_weight_wh
    if loss_weight_conf_target is not None and isinstance(
            loss_weight_conf_target, float):
        self.attrs['loss_weight_conf_target'] = loss_weight_conf_target
    if loss_weight_conf_notarget is not None and isinstance(
            loss_weight_conf_notarget, float):
        self.attrs['loss_weight_conf_notarget'] = loss_weight_conf_notarget
    if loss_weight_class is not None and isinstance(loss_weight_class, float):
        self.attrs['loss_weight_class'] = loss_weight_class
D
dengkaipeng 已提交
491 492 493

    helper.append_op(
        type='yolov3_loss',
D
dengkaipeng 已提交
494 495 496
        inputs={"X": x,
                "GTBox": gtbox,
                "GTLabel": gtlabel},
D
dengkaipeng 已提交
497 498 499 500 501
        outputs={'Loss': loss},
        attrs=attrs)
    return loss


X
Xin Pan 已提交
502
@templatedoc()
503 504
def detection_map(detect_res,
                  label,
505 506
                  class_num,
                  background_label=0,
507 508
                  overlap_threshold=0.3,
                  evaluate_difficult=True,
509 510 511 512
                  has_state=None,
                  input_states=None,
                  out_states=None,
                  ap_version='integral'):
X
Xin Pan 已提交
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553
    """
    ${comment}

    Args:
        detect_res: ${detect_res_comment}
        label:  ${label_comment}
        class_num: ${class_num_comment}
        background_label: ${background_label_comment}
        overlap_threshold: ${overlap_threshold_comment}
        evaluate_difficult: ${evaluate_difficult_comment}
        has_state: ${has_state_comment}
        input_states: If not None, It contains 3 elements:
            1. pos_count ${pos_count_comment}.
            2. true_pos ${true_pos_comment}.
            3. false_pos ${false_pos_comment}.
        out_states: If not None, it contains 3 elements.
            1. accum_pos_count ${accum_pos_count_comment}.
            2. accum_true_pos ${accum_true_pos_comment}.
            3. accum_false_pos ${accum_false_pos_comment}.
        ap_version: ${ap_type_comment}

    Returns:
        ${map_comment}


    Examples:
          .. code-block:: python

            detect_res = fluid.layers.data(
                name='detect_res',
                shape=[10, 6],
                append_batch_size=False,
                dtype='float32')
            label = fluid.layers.data(
                name='label',
                shape=[10, 6],
                append_batch_size=False,
                dtype='float32')

            map_out = fluid.layers.detection_map(detect_res, label, 21)
    """
554 555
    helper = LayerHelper("detection_map", **locals())

556
    def __create_var(type):
X
Xin Pan 已提交
557
        return helper.create_variable_for_type_inference(dtype=type)
558 559 560 561 562 563 564 565 566 567 568 569

    map_out = __create_var('float32')
    accum_pos_count_out = out_states[0] if out_states else __create_var('int32')
    accum_true_pos_out = out_states[1] if out_states else __create_var(
        'float32')
    accum_false_pos_out = out_states[2] if out_states else __create_var(
        'float32')

    pos_count = input_states[0] if input_states else None
    true_pos = input_states[1] if input_states else None
    false_pos = input_states[2] if input_states else None

570 571 572 573 574
    helper.append_op(
        type="detection_map",
        inputs={
            'Label': label,
            'DetectRes': detect_res,
575
            'HasState': has_state,
576 577 578 579 580 581 582 583 584 585 586 587 588
            'PosCount': pos_count,
            'TruePos': true_pos,
            'FalsePos': false_pos
        },
        outputs={
            'MAP': map_out,
            'AccumPosCount': accum_pos_count_out,
            'AccumTruePos': accum_true_pos_out,
            'AccumFalsePos': accum_false_pos_out
        },
        attrs={
            'overlap_threshold': overlap_threshold,
            'evaluate_difficult': evaluate_difficult,
589 590
            'ap_type': ap_version,
            'class_num': class_num,
591
        })
592
    return map_out
593 594


595 596 597 598
def bipartite_match(dist_matrix,
                    match_type=None,
                    dist_threshold=None,
                    name=None):
599
    """
Y
yuyang18 已提交
600 601
    This operator implements a greedy bipartite matching algorithm, which is
    used to obtain the matching with the maximum distance based on the input
602
    distance matrix. For input 2D matrix, the bipartite matching algorithm can
Y
yuyang18 已提交
603 604 605 606 607 608 609 610
    find the matched column for each row (matched means the largest distance),
    also can find the matched row for each column. And this operator only
    calculate matched indices from column to row. For each instance,
    the number of matched indices is the column number of the input distance
    matrix.

    There are two outputs, matched indices and distance.
    A simple description, this algorithm matched the best (maximum distance)
611 612 613
    row entity to the column entity and the matched indices are not duplicated
    in each row of ColToRowMatchIndices. If the column entity is not matched
    any row entity, set -1 in ColToRowMatchIndices.
C
chengduoZH 已提交
614

Y
yuyang18 已提交
615
    NOTE: the input DistMat can be LoDTensor (with LoD) or Tensor.
616 617 618
    If LoDTensor with LoD, the height of ColToRowMatchIndices is batch size.
    If Tensor, the height of ColToRowMatchIndices is 1.

Y
yuyang18 已提交
619 620 621
    NOTE: This API is a very low level API. It is used by :code:`ssd_loss`
    layer. Please consider to use :code:`ssd_loss` instead.

622 623 624 625 626
    Args:
        dist_matrix(Variable): This input is a 2-D LoDTensor with shape
            [K, M]. It is pair-wise distance matrix between the entities
            represented by each row and each column. For example, assumed one
            entity is A with shape [K], another entity is B with shape [M]. The
Y
yuyang18 已提交
627 628 629 630 631 632
            dist_matrix[i][j] is the distance between A[i] and B[j]. The bigger
            the distance is, the better matching the pairs are.

            NOTE: This tensor can contain LoD information to represent a batch
            of inputs. One instance of this batch can contain different numbers
            of entities.
633
        match_type(string|None): The type of matching method, should be
Y
yuyang18 已提交
634
           'bipartite' or 'per_prediction'. [default 'bipartite'].
635 636
        dist_threshold(float|None): If `match_type` is 'per_prediction',
            this threshold is to determine the extra matching bboxes based
Y
yuyang18 已提交
637
            on the maximum distance, 0.5 by default.
638
    Returns:
Y
yuyang18 已提交
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
        tuple: a tuple with two elements is returned. The first is
        matched_indices, the second is matched_distance.

        The matched_indices is a 2-D Tensor with shape [N, M] in int type.
        N is the batch size. If match_indices[i][j] is -1, it
        means B[j] does not match any entity in i-th instance.
        Otherwise, it means B[j] is matched to row
        match_indices[i][j] in i-th instance. The row number of
        i-th instance is saved in match_indices[i][j].

        The matched_distance is a 2-D Tensor with shape [N, M] in float type
        . N is batch size. If match_indices[i][j] is -1,
        match_distance[i][j] is also -1.0. Otherwise, assumed
        match_distance[i][j] = d, and the row offsets of each instance
        are called LoD. Then match_distance[i][j] =
        dist_matrix[d+LoD[i]][j].

    Examples:

        >>> x = fluid.layers.data(name='x', shape=[4], dtype='float32')
        >>> y = fluid.layers.data(name='y', shape=[4], dtype='float32')
        >>> iou = fluid.layers.iou_similarity(x=x, y=y)
        >>> matched_indices, matched_dist = fluid.layers.bipartite_match(iou)
662 663
    """
    helper = LayerHelper('bipartite_match', **locals())
X
Xin Pan 已提交
664 665 666
    match_indices = helper.create_variable_for_type_inference(dtype='int32')
    match_distance = helper.create_variable_for_type_inference(
        dtype=dist_matrix.dtype)
667 668 669
    helper.append_op(
        type='bipartite_match',
        inputs={'DistMat': dist_matrix},
670 671 672 673
        attrs={
            'match_type': match_type,
            'dist_threshold': dist_threshold,
        },
674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690
        outputs={
            'ColToRowMatchIndices': match_indices,
            'ColToRowMatchDist': match_distance
        })
    return match_indices, match_distance


def target_assign(input,
                  matched_indices,
                  negative_indices=None,
                  mismatch_value=None,
                  name=None):
    """
    This operator can be, for given the target bounding boxes or labels,
    to assign classification and regression targets to each prediction as well as
    weights to prediction. The weights is used to specify which prediction would
    not contribute to training loss.
C
chengduoZH 已提交
691

692 693 694 695 696
    For each instance, the output `out` and`out_weight` are assigned based on
    `match_indices` and `negative_indices`.
    Assumed that the row offset for each instance in `input` is called lod,
    this operator assigns classification/regression targets by performing the
    following steps:
C
chengduoZH 已提交
697

698
    1. Assigning all outpts based on `match_indices`:
C
chengduoZH 已提交
699

700 701 702
    .. code-block:: text

        If id = match_indices[i][j] > 0,
C
chengduoZH 已提交
703

704 705
            out[i][j][0 : K] = X[lod[i] + id][j % P][0 : K]
            out_weight[i][j] = 1.
C
chengduoZH 已提交
706

707
        Otherwise,
C
chengduoZH 已提交
708

709 710
            out[j][j][0 : K] = {mismatch_value, mismatch_value, ...}
            out_weight[i][j] = 0.
C
chengduoZH 已提交
711

712
    2. Assigning out_weight based on `neg_indices` if `neg_indices` is provided:
C
chengduoZH 已提交
713

714 715
    Assumed that the row offset for each instance in `neg_indices` is called neg_lod,
    for i-th instance and each `id` of neg_indices in this instance:
M
minqiyang 已提交
716

717
    .. code-block:: text
C
chengduoZH 已提交
718

719 720 721 722 723 724 725 726 727 728 729 730 731 732 733
        out[i][id][0 : K] = {mismatch_value, mismatch_value, ...}
        out_weight[i][id] = 1.0

    Args:
       inputs (Variable): This input is a 3D LoDTensor with shape [M, P, K].
       matched_indices (Variable): Tensor<int>), The input matched indices
           is 2D Tenosr<int32> with shape [N, P], If MatchIndices[i][j] is -1,
           the j-th entity of column is not matched to any entity of row in
           i-th instance.
       negative_indices (Variable): The input negative example indices are
           an optional input with shape [Neg, 1] and int32 type, where Neg is
           the total number of negative example indices.
       mismatch_value (float32): Fill this value to the mismatched location.

    Returns:
M
minqiyang 已提交
734 735 736 737 738
        tuple:
               A tuple(out, out_weight) is returned. out is a 3D Tensor with
               shape [N, P, K], N and P is the same as they are in
               `neg_indices`, K is the same as it in input of X. If
               `match_indices[i][j]`. out_weight is the weight for output with
739 740 741 742 743 744 745 746 747 748 749
               the shape of [N, P, 1].

    Examples:

        .. code-block:: python

            matched_indices, matched_dist = fluid.layers.bipartite_match(iou)
            gt = layers.data(
                        name='gt', shape=[1, 1], dtype='int32', lod_level=1)
            trg, trg_weight = layers.target_assign(
                            gt, matched_indices, mismatch_value=0)
750 751
    """
    helper = LayerHelper('target_assign', **locals())
X
Xin Pan 已提交
752 753
    out = helper.create_variable_for_type_inference(dtype=input.dtype)
    out_weight = helper.create_variable_for_type_inference(dtype='float32')
754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780
    helper.append_op(
        type='target_assign',
        inputs={
            'X': input,
            'MatchIndices': matched_indices,
            'NegIndices': negative_indices
        },
        outputs={'Out': out,
                 'OutWeight': out_weight},
        attrs={'mismatch_value': mismatch_value})
    return out, out_weight


def ssd_loss(location,
             confidence,
             gt_box,
             gt_label,
             prior_box,
             prior_box_var=None,
             background_label=0,
             overlap_threshold=0.5,
             neg_pos_ratio=3.0,
             neg_overlap=0.5,
             loc_loss_weight=1.0,
             conf_loss_weight=1.0,
             match_type='per_prediction',
             mining_type='max_negative',
781
             normalize=True,
782 783
             sample_size=None):
    """
Y
yuyang18 已提交
784
    **Multi-box loss layer for object detection algorithm of SSD**
785 786 787 788 789 790 791

    This layer is to compute dection loss for SSD given the location offset
    predictions, confidence predictions, prior boxes and ground-truth boudding
    boxes and labels, and the type of hard example mining. The returned loss
    is a weighted sum of the localization loss (or regression loss) and
    confidence loss (or classification loss) by performing the following steps:

Y
yuyang18 已提交
792
    1. Find matched bounding box by bipartite matching algorithm.
Y
yuyang18 已提交
793

794
      1.1 Compute IOU similarity between ground-truth boxes and prior boxes.
Y
yuyang18 已提交
795

796
      1.2 Compute matched boundding box by bipartite matching algorithm.
Y
yuyang18 已提交
797

798
    2. Compute confidence for mining hard examples
Y
yuyang18 已提交
799

800
      2.1. Get the target label based on matched indices.
Y
yuyang18 已提交
801

802
      2.2. Compute confidence loss.
Y
yuyang18 已提交
803

804 805
    3. Apply hard example mining to get the negative example indices and update
       the matched indices.
Y
yuyang18 已提交
806

807
    4. Assign classification and regression targets
Y
yuyang18 已提交
808

809
      4.1. Encoded bbox according to the prior boxes.
Y
yuyang18 已提交
810

811
      4.2. Assign regression targets.
Y
yuyang18 已提交
812

813
      4.3. Assign classification targets.
Y
yuyang18 已提交
814

815
    5. Compute the overall objective loss.
Y
yuyang18 已提交
816

817
      5.1 Compute confidence loss.
Y
yuyang18 已提交
818

819
      5.1 Compute localization loss.
Y
yuyang18 已提交
820

821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843
      5.3 Compute the overall weighted loss.

    Args:
        location (Variable): The location predictions are a 3D Tensor with
            shape [N, Np, 4], N is the batch size, Np is total number of
            predictions for each instance. 4 is the number of coordinate values,
            the layout is [xmin, ymin, xmax, ymax].
        confidence (Variable): The confidence predictions are a 3D Tensor
            with shape [N, Np, C], N and Np are the same as they are in
            `location`, C is the class number.
        gt_box (Variable): The ground-truth boudding boxes (bboxes) are a 2D
            LoDTensor with shape [Ng, 4], Ng is the total number of ground-truth
            bboxes of mini-batch input.
        gt_label (Variable): The ground-truth labels are a 2D LoDTensor
            with shape [Ng, 1].
        prior_box (Variable): The prior boxes are a 2D Tensor with shape [Np, 4].
        prior_box_var (Variable): The variance of prior boxes are a 2D Tensor
            with shape [Np, 4].
        background_label (int): The index of background label, 0 by default.
        overlap_threshold (float): If match_type is 'per_prediction', use
            `overlap_threshold` to determine the extra matching bboxes when
             finding matched boxes. 0.5 by default.
        neg_pos_ratio (float): The ratio of the negative boxes to the positive
844
            boxes, used only when mining_type is 'max_negative', 3.0 by defalut.
845
        neg_overlap (float): The negative overlap upper bound for the unmatched
846
            predictions. Use only when mining_type is 'max_negative',
847 848 849 850
            0.5 by default.
        loc_loss_weight (float): Weight for localization loss, 1.0 by default.
        conf_loss_weight (float): Weight for confidence loss, 1.0 by default.
        match_type (str): The type of matching method during training, should
851
            be 'bipartite' or 'per_prediction', 'per_prediction' by defalut.
852 853
        mining_type (str): The hard example mining type, should be 'hard_example'
            or 'max_negative', now only support `max_negative`.
854
        normalize (bool): Whether to normalize the SSD loss by the total number
Y
yuyang18 已提交
855
            of output locations, True by default.
856 857
        sample_size (int): The max sample size of negative box, used only when
            mining_type is 'hard_example'.
858 859

    Returns:
Y
yuyang18 已提交
860 861
        The weighted sum of the localization loss and confidence loss, with \
        shape [N * Np, 1], N and Np are the same as they are in `location`.
862 863

    Raises:
Y
yuyang18 已提交
864 865
        ValueError: If mining_type is 'hard_example', now only support mining \
        type of `max_negative`.
Y
yuyang18 已提交
866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884

    Examples:
        >>> pb = fluid.layers.data(
        >>>                   name='prior_box',
        >>>                   shape=[10, 4],
        >>>                   append_batch_size=False,
        >>>                   dtype='float32')
        >>> pbv = fluid.layers.data(
        >>>                   name='prior_box_var',
        >>>                   shape=[10, 4],
        >>>                   append_batch_size=False,
        >>>                   dtype='float32')
        >>> loc = fluid.layers.data(name='target_box', shape=[10, 4], dtype='float32')
        >>> scores = fluid.layers.data(name='scores', shape=[10, 21], dtype='float32')
        >>> gt_box = fluid.layers.data(
        >>>         name='gt_box', shape=[4], lod_level=1, dtype='float32')
        >>> gt_label = fluid.layers.data(
        >>>         name='gt_label', shape=[1], lod_level=1, dtype='float32')
        >>> loss = fluid.layers.ssd_loss(loc, scores, gt_box, gt_label, pb, pbv)
885 886 887 888 889 890 891
    """

    helper = LayerHelper('ssd_loss', **locals())
    if mining_type != 'max_negative':
        raise ValueError("Only support mining_type == max_negative now.")

    num, num_prior, num_class = confidence.shape
G
merge  
gongweibao 已提交
892
    conf_shape = nn.shape(confidence)
893 894

    def __reshape_to_2d(var):
895
        return nn.flatten(x=var, axis=2)
896 897 898 899 900

    # 1. Find matched boundding box by prior box.
    #   1.1 Compute IOU similarity between ground-truth boxes and prior boxes.
    iou = iou_similarity(x=gt_box, y=prior_box)
    #   1.2 Compute matched boundding box by bipartite matching algorithm.
901 902
    matched_indices, matched_dist = bipartite_match(iou, match_type,
                                                    overlap_threshold)
903 904 905

    # 2. Compute confidence for mining hard examples
    # 2.1. Get the target label based on matched indices
906 907
    gt_label = nn.reshape(
        x=gt_label, shape=(len(gt_label.shape) - 1) * (0, ) + (-1, 1))
908
    gt_label.stop_gradient = True
909 910 911 912 913 914 915
    target_label, _ = target_assign(
        gt_label, matched_indices, mismatch_value=background_label)
    # 2.2. Compute confidence loss.
    # Reshape confidence to 2D tensor.
    confidence = __reshape_to_2d(confidence)
    target_label = tensor.cast(x=target_label, dtype='int64')
    target_label = __reshape_to_2d(target_label)
916
    target_label.stop_gradient = True
917 918
    conf_loss = nn.softmax_with_cross_entropy(confidence, target_label)
    # 3. Mining hard examples
G
merge  
gongweibao 已提交
919
    actual_shape = nn.slice(conf_shape, axes=[0], starts=[0], ends=[2])
920
    actual_shape.stop_gradient = True
921
    conf_loss = nn.reshape(
922
        x=conf_loss, shape=(num, num_prior), actual_shape=actual_shape)
923
    conf_loss.stop_gradient = True
X
Xin Pan 已提交
924
    neg_indices = helper.create_variable_for_type_inference(dtype='int32')
925
    dtype = matched_indices.dtype
X
Xin Pan 已提交
926 927
    updated_matched_indices = helper.create_variable_for_type_inference(
        dtype=dtype)
928 929 930 931 932 933 934 935 936 937 938 939 940 941
    helper.append_op(
        type='mine_hard_examples',
        inputs={
            'ClsLoss': conf_loss,
            'LocLoss': None,
            'MatchIndices': matched_indices,
            'MatchDist': matched_dist,
        },
        outputs={
            'NegIndices': neg_indices,
            'UpdatedMatchIndices': updated_matched_indices
        },
        attrs={
            'neg_pos_ratio': neg_pos_ratio,
B
Bai Yifan 已提交
942
            'neg_dist_threshold': neg_overlap,
943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967
            'mining_type': mining_type,
            'sample_size': sample_size,
        })

    # 4. Assign classification and regression targets
    # 4.1. Encoded bbox according to the prior boxes.
    encoded_bbox = box_coder(
        prior_box=prior_box,
        prior_box_var=prior_box_var,
        target_box=gt_box,
        code_type='encode_center_size')
    # 4.2. Assign regression targets
    target_bbox, target_loc_weight = target_assign(
        encoded_bbox, updated_matched_indices, mismatch_value=background_label)
    # 4.3. Assign classification targets
    target_label, target_conf_weight = target_assign(
        gt_label,
        updated_matched_indices,
        negative_indices=neg_indices,
        mismatch_value=background_label)

    # 5. Compute loss.
    # 5.1 Compute confidence loss.
    target_label = __reshape_to_2d(target_label)
    target_label = tensor.cast(x=target_label, dtype='int64')
968

969 970 971 972
    conf_loss = nn.softmax_with_cross_entropy(confidence, target_label)
    target_conf_weight = __reshape_to_2d(target_conf_weight)
    conf_loss = conf_loss * target_conf_weight

973 974 975 976
    # the target_label and target_conf_weight do not have gradient.
    target_label.stop_gradient = True
    target_conf_weight.stop_gradient = True

977 978 979 980 981 982 983 984
    # 5.2 Compute regression loss.
    location = __reshape_to_2d(location)
    target_bbox = __reshape_to_2d(target_bbox)

    loc_loss = nn.smooth_l1(location, target_bbox)
    target_loc_weight = __reshape_to_2d(target_loc_weight)
    loc_loss = loc_loss * target_loc_weight

985 986 987 988
    # the target_bbox and target_loc_weight do not have gradient.
    target_bbox.stop_gradient = True
    target_loc_weight.stop_gradient = True

989 990
    # 5.3 Compute overall weighted loss.
    loss = conf_loss_weight * conf_loss + loc_loss_weight * loc_loss
991
    # reshape to [N, Np], N is the batch size and Np is the prior box number.
992
    loss = nn.reshape(x=loss, shape=(num, num_prior), actual_shape=actual_shape)
993 994 995 996 997
    loss = nn.reduce_sum(loss, dim=1, keep_dim=True)
    if normalize:
        normalizer = nn.reduce_sum(target_loc_weight)
        loss = loss / normalizer

998
    return loss
C
chengduoZH 已提交
999 1000


1001 1002 1003 1004
def prior_box(input,
              image,
              min_sizes,
              max_sizes=None,
1005
              aspect_ratios=[1.],
1006 1007 1008 1009 1010
              variance=[0.1, 0.1, 0.2, 0.2],
              flip=False,
              clip=False,
              steps=[0.0, 0.0],
              offset=0.5,
1011 1012
              name=None,
              min_max_aspect_ratios_order=False):
1013
    """
Q
update  
qiaolongfei 已提交
1014
    **Prior Box Operator**
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025

    Generate prior boxes for SSD(Single Shot MultiBox Detector) algorithm.
    Each position of the input produce N prior boxes, N is determined by
    the count of min_sizes, max_sizes and aspect_ratios, The size of the
    box is in range(min_size, max_size) interval, which is generated in
    sequence according to the aspect_ratios.

    Args:
       input(Variable): The Input Variables, the format is NCHW.
       image(Variable): The input image data of PriorBoxOp,
            the layout is NCHW.
1026
       min_sizes(list|tuple|float value): min sizes of generated prior boxes.
1027 1028
       max_sizes(list|tuple|None): max sizes of generated prior boxes.
            Default: None.
1029 1030
       aspect_ratios(list|tuple|float value): the aspect ratios of generated
            prior boxes. Default: [1.].
1031 1032 1033 1034
       variance(list|tuple): the variances to be encoded in prior boxes.
            Default:[0.1, 0.1, 0.2, 0.2].
       flip(bool): Whether to flip aspect ratios. Default:False.
       clip(bool): Whether to clip out-of-boundary boxes. Default: False.
1035
       step(list|turple): Prior boxes step across width and height, If
1036
            step[0] == 0.0/step[1] == 0.0, the prior boxes step across
1037 1038
            height/weight of the input will be automatically calculated.
            Default: [0., 0.]
1039 1040
       offset(float): Prior boxes center offset. Default: 0.5
       name(str): Name of the prior box op. Default: None.
1041
       min_max_aspect_ratios_order(bool): If set True, the output prior box is
M
minqiyang 已提交
1042
            in order of [min, max, aspect_ratios], which is consistent with
1043 1044 1045
            Caffe. Please note, this order affects the weights order of
            convolution layer followed by and does not affect the final
            detection results. Default: False.
1046 1047

    Returns:
Q
update  
qiaolongfei 已提交
1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
        tuple: A tuple with two Variable (boxes, variances)

        boxes: the output prior boxes of PriorBox.
        The layout is [H, W, num_priors, 4].
        H is the height of input, W is the width of input,
        num_priors is the total
        box count of each position of input.

        variances: the expanded variances of PriorBox.
        The layout is [H, W, num_priors, 4].
        H is the height of input, W is the width of input
        num_priors is the total
        box count of each position of input
1061 1062 1063 1064


    Examples:
        .. code-block:: python
Q
update  
qiaolongfei 已提交
1065 1066 1067 1068 1069 1070 1071

            box, var = fluid.layers.prior_box(
                input=conv1,
                image=images,
                min_sizes=[100.],
                flip=True,
                clip=True)
1072 1073 1074 1075
    """
    helper = LayerHelper("prior_box", **locals())
    dtype = helper.input_dtype()

1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
    def _is_list_or_tuple_(data):
        return (isinstance(data, list) or isinstance(data, tuple))

    if not _is_list_or_tuple_(min_sizes):
        min_sizes = [min_sizes]
    if not _is_list_or_tuple_(aspect_ratios):
        aspect_ratios = [aspect_ratios]
    if not (_is_list_or_tuple_(steps) and len(steps) == 2):
        raise ValueError('steps should be a list or tuple ',
                         'with length 2, (step_width, step_height).')

    min_sizes = list(map(float, min_sizes))
    aspect_ratios = list(map(float, aspect_ratios))
    steps = list(map(float, steps))

1091 1092 1093 1094 1095 1096 1097 1098
    attrs = {
        'min_sizes': min_sizes,
        'aspect_ratios': aspect_ratios,
        'variances': variance,
        'flip': flip,
        'clip': clip,
        'step_w': steps[0],
        'step_h': steps[1],
1099 1100
        'offset': offset,
        'min_max_aspect_ratios_order': min_max_aspect_ratios_order
1101 1102
    }
    if max_sizes is not None and len(max_sizes) > 0 and max_sizes[0] > 0:
1103 1104
        if not _is_list_or_tuple_(max_sizes):
            max_sizes = [max_sizes]
1105 1106
        attrs['max_sizes'] = max_sizes

X
Xin Pan 已提交
1107 1108
    box = helper.create_variable_for_type_inference(dtype)
    var = helper.create_variable_for_type_inference(dtype)
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
    helper.append_op(
        type="prior_box",
        inputs={"Input": input,
                "Image": image},
        outputs={"Boxes": box,
                 "Variances": var},
        attrs=attrs, )
    box.stop_gradient = True
    var.stop_gradient = True
    return box, var


R
ruri 已提交
1121 1122 1123 1124 1125 1126 1127 1128 1129
def density_prior_box(input,
                      image,
                      densities=None,
                      fixed_sizes=None,
                      fixed_ratios=None,
                      variance=[0.1, 0.1, 0.2, 0.2],
                      clip=False,
                      steps=[0.0, 0.0],
                      offset=0.5,
1130
                      flatten_to_2d=False,
R
ruri 已提交
1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
                      name=None):
    """
    **Density Prior Box Operator**

    Generate density prior boxes for SSD(Single Shot MultiBox Detector) 
    algorithm. Each position of the input produce N prior boxes, N is 
    determined by the count of densities, fixed_sizes and fixed_ratios. 
    Boxes center at grid points around each input position is generated by 
    this operator, and the grid points is determined by densities and 
    the count of density prior box is determined by fixed_sizes and fixed_ratios. 
    Obviously, the number of fixed_sizes is equal to the number of densities.
    For densities_i in densities:
    N_density_prior_box =sum(N_fixed_ratios * densities_i^2),

    Args:
       input(Variable): The Input Variables, the format is NCHW.
       image(Variable): The input image data of PriorBoxOp,
            the layout is NCHW.
       densities(list|tuple|None): the densities of generated density prior 
            boxes, this attribute should be a list or tuple of integers. 
            Default: None.
       fixed_sizes(list|tuple|None): the fixed sizes of generated density
            prior boxes, this attribute should a list or tuple of same 
            length with :attr:`densities`. Default: None.
       fixed_ratios(list|tuple|None): the fixed ratios of generated density
            prior boxes, if this attribute is not set and :attr:`densities`
            and :attr:`fix_sizes` is set, :attr:`aspect_ratios` will be used
            to generate density prior boxes.
       variance(list|tuple): the variances to be encoded in density prior boxes.
            Default:[0.1, 0.1, 0.2, 0.2].
       clip(bool): Whether to clip out-of-boundary boxes. Default: False.
       step(list|turple): Prior boxes step across width and height, If
            step[0] == 0.0/step[1] == 0.0, the density prior boxes step across
            height/weight of the input will be automatically calculated.
            Default: [0., 0.]
       offset(float): Prior boxes center offset. Default: 0.5
1167 1168
       flatten_to_2d(bool): Whether to flatten output prior boxes and variance
           to 2D shape, the second dim is 4. Default: False.
R
ruri 已提交
1169 1170 1171 1172 1173 1174
       name(str): Name of the density prior box op. Default: None.

    Returns:
        tuple: A tuple with two Variable (boxes, variances)

        boxes: the output density prior boxes of PriorBox.
1175 1176 1177 1178
            The layout is [H, W, num_priors, 4] when flatten_to_2d is False.
            The layout is [H * W * num_priors, 4] when flatten_to_2d is True.
            H is the height of input, W is the width of input,
            num_priors is the total box count of each position of input.
R
ruri 已提交
1179 1180

        variances: the expanded variances of PriorBox.
1181 1182 1183 1184
            The layout is [H, W, num_priors, 4] when flatten_to_2d is False.
            The layout is [H * W * num_priors, 4] when flatten_to_2d is True.
            H is the height of input, W is the width of input
            num_priors is the total box count of each position of input.
R
ruri 已提交
1185 1186 1187 1188 1189 1190 1191 1192


    Examples:
        .. code-block:: python

            box, var = fluid.layers.density_prior_box(
                input=conv1,
                image=images,
1193 1194 1195 1196 1197
                densities=[4, 2, 1],
                fixed_sizes=[32.0, 64.0, 128.0],
                fixed_ratios=[1.],
                clip=True,
                flatten_to_2d=True)
R
ruri 已提交
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
    """
    helper = LayerHelper("density_prior_box", **locals())
    dtype = helper.input_dtype()

    def _is_list_or_tuple_(data):
        return (isinstance(data, list) or isinstance(data, tuple))

    if not _is_list_or_tuple_(densities):
        raise TypeError('densities should be a list or a tuple or None.')
    if not _is_list_or_tuple_(fixed_sizes):
        raise TypeError('fixed_sizes should be a list or a tuple or None.')
    if not _is_list_or_tuple_(fixed_ratios):
        raise TypeError('fixed_ratios should be a list or a tuple or None.')
    if len(densities) != len(fixed_sizes):
        raise ValueError('densities and fixed_sizes length should be euqal.')
    if not (_is_list_or_tuple_(steps) and len(steps) == 2):
        raise ValueError('steps should be a list or tuple ',
                         'with length 2, (step_width, step_height).')

    densities = list(map(int, densities))
    fixed_sizes = list(map(float, fixed_sizes))
    fixed_ratios = list(map(float, fixed_ratios))
    steps = list(map(float, steps))

    attrs = {
        'variances': variance,
        'clip': clip,
        'step_w': steps[0],
        'step_h': steps[1],
        'offset': offset,
1228 1229 1230 1231
        'densities': densities,
        'fixed_sizes': fixed_sizes,
        'fixed_ratios': fixed_ratios,
        'flatten_to_2d': flatten_to_2d,
R
ruri 已提交
1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
    }
    box = helper.create_variable_for_type_inference(dtype)
    var = helper.create_variable_for_type_inference(dtype)
    helper.append_op(
        type="density_prior_box",
        inputs={"Input": input,
                "Image": image},
        outputs={"Boxes": box,
                 "Variances": var},
        attrs=attrs, )
    box.stop_gradient = True
    var.stop_gradient = True
    return box, var


C
chengduoZH 已提交
1247
def multi_box_head(inputs,
C
chengduoZH 已提交
1248 1249
                   image,
                   base_size,
C
chengduoZH 已提交
1250
                   num_classes,
C
chengduoZH 已提交
1251
                   aspect_ratios,
1252 1253
                   min_ratio=None,
                   max_ratio=None,
C
chengduoZH 已提交
1254 1255
                   min_sizes=None,
                   max_sizes=None,
C
chengduoZH 已提交
1256 1257 1258 1259
                   steps=None,
                   step_w=None,
                   step_h=None,
                   offset=0.5,
1260 1261
                   variance=[0.1, 0.1, 0.2, 0.2],
                   flip=True,
C
chengduoZH 已提交
1262
                   clip=False,
C
chengduoZH 已提交
1263
                   kernel_size=1,
C
chengduoZH 已提交
1264
                   pad=0,
C
chengduoZH 已提交
1265
                   stride=1,
1266 1267
                   name=None,
                   min_max_aspect_ratios_order=False):
C
chengduoZH 已提交
1268
    """
C
chengduoZH 已提交
1269 1270
    Generate prior boxes for SSD(Single Shot MultiBox Detector)
    algorithm. The details of this algorithm, please refer the
Q
update  
qiaolongfei 已提交
1271
    section 2.2 of SSD paper `SSD: Single Shot MultiBox Detector
C
chengduoZH 已提交
1272
    <https://arxiv.org/abs/1512.02325>`_ .
C
chengduoZH 已提交
1273 1274

    Args:
1275
       inputs(list|tuple): The list of input Variables, the format
C
chengduoZH 已提交
1276
            of all Variables is NCHW.
C
chengduoZH 已提交
1277 1278
       image(Variable): The input image data of PriorBoxOp,
            the layout is NCHW.
C
chengduoZH 已提交
1279 1280
       base_size(int): the base_size is used to get min_size
            and max_size according to min_ratio and max_ratio.
C
chengduoZH 已提交
1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302
       num_classes(int): The number of classes.
       aspect_ratios(list|tuple): the aspect ratios of generated prior
            boxes. The length of input and aspect_ratios must be equal.
       min_ratio(int): the min ratio of generated prior boxes.
       max_ratio(int): the max ratio of generated prior boxes.
       min_sizes(list|tuple|None): If `len(inputs) <=2`,
            min_sizes must be set up, and the length of min_sizes
            should equal to the length of inputs. Default: None.
       max_sizes(list|tuple|None): If `len(inputs) <=2`,
            max_sizes must be set up, and the length of min_sizes
            should equal to the length of inputs. Default: None.
       steps(list|tuple): If step_w and step_h are the same,
            step_w and step_h can be replaced by steps.
       step_w(list|tuple): Prior boxes step
            across width. If step_w[i] == 0.0, the prior boxes step
            across width of the inputs[i] will be automatically
            calculated. Default: None.
       step_h(list|tuple): Prior boxes step across height, If
            step_h[i] == 0.0, the prior boxes step across height of
            the inputs[i] will be automatically calculated. Default: None.
       offset(float): Prior boxes center offset. Default: 0.5
       variance(list|tuple): the variances to be encoded in prior boxes.
1303
            Default:[0.1, 0.1, 0.2, 0.2].
C
chengduoZH 已提交
1304 1305 1306 1307 1308 1309
       flip(bool): Whether to flip aspect ratios. Default:False.
       clip(bool): Whether to clip out-of-boundary boxes. Default: False.
       kernel_size(int): The kernel size of conv2d. Default: 1.
       pad(int|list|tuple): The padding of conv2d. Default:0.
       stride(int|list|tuple): The stride of conv2d. Default:1,
       name(str): Name of the prior box layer. Default: None.
1310
       min_max_aspect_ratios_order(bool): If set True, the output prior box is
M
minqiyang 已提交
1311
            in order of [min, max, aspect_ratios], which is consistent with
1312 1313 1314
            Caffe. Please note, this order affects the weights order of
            convolution layer followed by and does not affect the fininal
            detection results. Default: False.
C
chengduoZH 已提交
1315 1316

    Returns:
Q
update  
qiaolongfei 已提交
1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331
        tuple: A tuple with four Variables. (mbox_loc, mbox_conf, boxes, variances)

        mbox_loc: The predicted boxes' location of the inputs. The layout
        is [N, H*W*Priors, 4]. where Priors is the number of predicted
        boxes each position of each input.

        mbox_conf: The predicted boxes' confidence of the inputs. The layout
        is [N, H*W*Priors, C]. where Priors is the number of predicted boxes
        each position of each input and C is the number of Classes.

        boxes: the output prior boxes of PriorBox. The layout is [num_priors, 4].
        num_priors is the total box count of each position of inputs.

        variances: the expanded variances of PriorBox. The layout is
        [num_priors, 4]. num_priors is the total box count of each position of inputs
C
chengduoZH 已提交
1332

C
chengduoZH 已提交
1333 1334 1335

    Examples:
        .. code-block:: python
Q
update  
qiaolongfei 已提交
1336 1337

          mbox_locs, mbox_confs, box, var = fluid.layers.multi_box_head(
C
chengduoZH 已提交
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347
            inputs=[conv1, conv2, conv3, conv4, conv5, conv5],
            image=images,
            num_classes=21,
            min_ratio=20,
            max_ratio=90,
            aspect_ratios=[[2.], [2., 3.], [2., 3.], [2., 3.], [2.], [2.]],
            base_size=300,
            offset=0.5,
            flip=True,
            clip=True)
C
chengduoZH 已提交
1348 1349
    """

C
chengduoZH 已提交
1350
    def _reshape_with_axis_(input, axis=1):
1351
        out = nn.flatten(x=input, axis=axis)
C
chengduoZH 已提交
1352
        return out
1353

1354 1355
    def _is_list_or_tuple_(data):
        return (isinstance(data, list) or isinstance(data, tuple))
1356

C
chengduoZH 已提交
1357 1358 1359 1360
    def _is_list_or_tuple_and_equal(data, length, err_info):
        if not (_is_list_or_tuple_(data) and len(data) == length):
            raise ValueError(err_info)

1361 1362
    if not _is_list_or_tuple_(inputs):
        raise ValueError('inputs should be a list or tuple.')
C
chengduoZH 已提交
1363

C
chengduoZH 已提交
1364 1365 1366 1367 1368
    num_layer = len(inputs)

    if num_layer <= 2:
        assert min_sizes is not None and max_sizes is not None
        assert len(min_sizes) == num_layer and len(max_sizes) == num_layer
1369
    elif min_sizes is None and max_sizes is None:
C
chengduoZH 已提交
1370 1371 1372
        min_sizes = []
        max_sizes = []
        step = int(math.floor(((max_ratio - min_ratio)) / (num_layer - 2)))
M
minqiyang 已提交
1373
        for ratio in six.moves.range(min_ratio, max_ratio + 1, step):
C
chengduoZH 已提交
1374 1375 1376 1377 1378
            min_sizes.append(base_size * ratio / 100.)
            max_sizes.append(base_size * (ratio + step) / 100.)
        min_sizes = [base_size * .10] + min_sizes
        max_sizes = [base_size * .20] + max_sizes

C
chengduoZH 已提交
1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
    if aspect_ratios:
        _is_list_or_tuple_and_equal(
            aspect_ratios, num_layer,
            'aspect_ratios should be list or tuple, and the length of inputs '
            'and aspect_ratios should be the same.')
    if step_h:
        _is_list_or_tuple_and_equal(
            step_h, num_layer,
            'step_h should be list or tuple, and the length of inputs and '
            'step_h should be the same.')
    if step_w:
        _is_list_or_tuple_and_equal(
            step_w, num_layer,
            'step_w should be list or tuple, and the length of inputs and '
            'step_w should be the same.')
    if steps:
        _is_list_or_tuple_and_equal(
            steps, num_layer,
            'steps should be list or tuple, and the length of inputs and '
            'step_w should be the same.')
        step_w = steps
        step_h = steps

C
chengduoZH 已提交
1402 1403
    mbox_locs = []
    mbox_confs = []
C
chengduoZH 已提交
1404 1405
    box_results = []
    var_results = []
C
chengduoZH 已提交
1406 1407
    for i, input in enumerate(inputs):
        min_size = min_sizes[i]
C
chengduoZH 已提交
1408 1409
        max_size = max_sizes[i]

1410
        if not _is_list_or_tuple_(min_size):
C
chengduoZH 已提交
1411
            min_size = [min_size]
C
chengduoZH 已提交
1412 1413
        if not _is_list_or_tuple_(max_size):
            max_size = [max_size]
C
chengduoZH 已提交
1414 1415 1416 1417

        aspect_ratio = []
        if aspect_ratios is not None:
            aspect_ratio = aspect_ratios[i]
1418
            if not _is_list_or_tuple_(aspect_ratio):
C
chengduoZH 已提交
1419
                aspect_ratio = [aspect_ratio]
1420
        step = [step_w[i] if step_w else 0.0, step_h[i] if step_w else 0.0]
C
chengduoZH 已提交
1421

1422
        box, var = prior_box(input, image, min_size, max_size, aspect_ratio,
1423 1424
                             variance, flip, clip, step, offset, None,
                             min_max_aspect_ratios_order)
C
chengduoZH 已提交
1425 1426 1427 1428 1429

        box_results.append(box)
        var_results.append(var)

        num_boxes = box.shape[2]
C
chengduoZH 已提交
1430

1431
        # get loc
Y
Yuan Gao 已提交
1432
        num_loc_output = num_boxes * 4
1433
        mbox_loc = nn.conv2d(
C
chengduoZH 已提交
1434
            input=input,
1435 1436 1437 1438 1439
            num_filters=num_loc_output,
            filter_size=kernel_size,
            padding=pad,
            stride=stride)

1440
        mbox_loc = nn.transpose(mbox_loc, perm=[0, 2, 3, 1])
1441
        compile_shape = [
1442
            mbox_loc.shape[0], cpt.floor_division(
M
minqiyang 已提交
1443
                mbox_loc.shape[1] * mbox_loc.shape[2] * mbox_loc.shape[3], 4), 4
Y
Yuan Gao 已提交
1444
        ]
1445 1446 1447
        run_shape = tensor.assign(numpy.array([0, -1, 4]).astype("int32"))
        mbox_loc_flatten = nn.reshape(
            mbox_loc, shape=compile_shape, actual_shape=run_shape)
Y
Yuan Gao 已提交
1448
        mbox_locs.append(mbox_loc_flatten)
C
chengduoZH 已提交
1449

1450
        # get conf
C
chengduoZH 已提交
1451
        num_conf_output = num_boxes * num_classes
1452
        conf_loc = nn.conv2d(
C
chengduoZH 已提交
1453
            input=input,
1454 1455 1456 1457
            num_filters=num_conf_output,
            filter_size=kernel_size,
            padding=pad,
            stride=stride)
1458
        conf_loc = nn.transpose(conf_loc, perm=[0, 2, 3, 1])
1459 1460
        new_shape = [0, -1, num_classes]
        compile_shape = [
1461 1462 1463
            conf_loc.shape[0],
            cpt.floor_division(conf_loc.shape[1] * conf_loc.shape[2] *
                               conf_loc.shape[3], num_classes), num_classes
Y
Yuan Gao 已提交
1464
        ]
1465 1466 1467 1468
        run_shape = tensor.assign(
            numpy.array([0, -1, num_classes]).astype("int32"))
        conf_loc_flatten = nn.reshape(
            conf_loc, shape=compile_shape, actual_shape=run_shape)
Y
Yuan Gao 已提交
1469
        mbox_confs.append(conf_loc_flatten)
C
chengduoZH 已提交
1470

C
chengduoZH 已提交
1471 1472 1473
    if len(box_results) == 1:
        box = box_results[0]
        var = var_results[0]
Y
Yuan Gao 已提交
1474 1475
        mbox_locs_concat = mbox_locs[0]
        mbox_confs_concat = mbox_confs[0]
C
chengduoZH 已提交
1476 1477 1478 1479 1480 1481 1482 1483 1484
    else:
        reshaped_boxes = []
        reshaped_vars = []
        for i in range(len(box_results)):
            reshaped_boxes.append(_reshape_with_axis_(box_results[i], axis=3))
            reshaped_vars.append(_reshape_with_axis_(var_results[i], axis=3))

        box = tensor.concat(reshaped_boxes)
        var = tensor.concat(reshaped_vars)
Y
Yuan Gao 已提交
1485 1486
        mbox_locs_concat = tensor.concat(mbox_locs, axis=1)
        mbox_confs_concat = tensor.concat(mbox_confs, axis=1)
C
chengduoZH 已提交
1487

1488 1489
    box.stop_gradient = True
    var.stop_gradient = True
Y
Yuan Gao 已提交
1490
    return mbox_locs_concat, mbox_confs_concat, box, var
1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510


def anchor_generator(input,
                     anchor_sizes=None,
                     aspect_ratios=None,
                     variance=[0.1, 0.1, 0.2, 0.2],
                     stride=None,
                     offset=0.5,
                     name=None):
    """
    **Anchor generator operator**

    Generate anchors for Faster RCNN algorithm.
    Each position of the input produce N anchors, N =
    size(anchor_sizes) * size(aspect_ratios). The order of generated anchors
    is firstly aspect_ratios loop then anchor_sizes loop.

    Args:
       input(Variable): The input feature map, the format is NCHW.
       anchor_sizes(list|tuple|float): The anchor sizes of generated anchors,
H
haowang101779990 已提交
1511 1512
                                       given in absolute pixels e.g. [64., 128., 256., 512.].
                                       For instance, the anchor size of 64 means the area of this anchor equals to 64**2.
1513
       aspect_ratios(list|tuple|float): The height / width ratios of generated
H
haowang101779990 已提交
1514
                                        anchors, e.g. [0.5, 1.0, 2.0].
1515
       variance(list|tuple): The variances to be used in box regression deltas.
H
haowang101779990 已提交
1516 1517
                             Default:[0.1, 0.1, 0.2, 0.2].
       stride(list|turple): The anchors stride across width and height,e.g. [16.0, 16.0]
1518 1519 1520 1521
       offset(float): Prior boxes center offset. Default: 0.5
       name(str): Name of the prior box op. Default: None.

    Returns:
H
haowang101779990 已提交
1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534
        Anchors(Variable),Variances(Variable):  
        
              two variables:
        
              - Anchors(Variable): The output anchors with a layout of [H, W, num_anchors, 4]. \
                H is the height of input, W is the width of input, \
                num_anchors is the box count of each position.  \
                Each anchor is in (xmin, ymin, xmax, ymax) format an unnormalized. 
              - Variances(Variable): The expanded variances of anchors \
                with a layout of [H, W, num_priors, 4]. \
                H is the height of input, W is the width of input \
                num_anchors is the box count of each position. \
                Each variance is in (xcenter, ycenter, w, h) format.
1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574


    Examples:

        .. code-block:: python

            anchor, var = anchor_generator(
                input=conv1,
                anchor_sizes=[64, 128, 256, 512],
                aspect_ratios=[0.5, 1.0, 2.0],
                variance=[0.1, 0.1, 0.2, 0.2],
                stride=[16.0, 16.0],
                offset=0.5)
    """
    helper = LayerHelper("anchor_generator", **locals())
    dtype = helper.input_dtype()

    def _is_list_or_tuple_(data):
        return (isinstance(data, list) or isinstance(data, tuple))

    if not _is_list_or_tuple_(anchor_sizes):
        anchor_sizes = [anchor_sizes]
    if not _is_list_or_tuple_(aspect_ratios):
        aspect_ratios = [aspect_ratios]
    if not (_is_list_or_tuple_(stride) and len(stride) == 2):
        raise ValueError('stride should be a list or tuple ',
                         'with length 2, (stride_width, stride_height).')

    anchor_sizes = list(map(float, anchor_sizes))
    aspect_ratios = list(map(float, aspect_ratios))
    stride = list(map(float, stride))

    attrs = {
        'anchor_sizes': anchor_sizes,
        'aspect_ratios': aspect_ratios,
        'variances': variance,
        'stride': stride,
        'offset': offset
    }

X
Xin Pan 已提交
1575 1576
    anchor = helper.create_variable_for_type_inference(dtype)
    var = helper.create_variable_for_type_inference(dtype)
1577 1578 1579 1580 1581 1582 1583 1584 1585
    helper.append_op(
        type="anchor_generator",
        inputs={"Input": input},
        outputs={"Anchors": anchor,
                 "Variances": var},
        attrs=attrs, )
    anchor.stop_gradient = True
    var.stop_gradient = True
    return anchor, var
1586 1587


W
whs 已提交
1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
def roi_perspective_transform(input,
                              rois,
                              transformed_height,
                              transformed_width,
                              spatial_scale=1.0):
    """
    ROI perspective transform op.

    Args:
        input (Variable): The input of ROIPerspectiveTransformOp. The format of 
                          input tensor is NCHW. Where N is batch size, C is the
                          number of input channels, H is the height of the feature,
                          and W is the width of the feature.
        rois (Variable):  ROIs (Regions of Interest) to be transformed. It should be
                          a 2-D LoDTensor of shape (num_rois, 8). Given as 
                          [[x1, y1, x2, y2, x3, y3, x4, y4], ...], (x1, y1) is the 
                          top left coordinates, and (x2, y2) is the top right 
                          coordinates, and (x3, y3) is the bottom right coordinates, 
                          and (x4, y4) is the bottom left coordinates.
        transformed_height (integer): The height of transformed output.
        transformed_height (integer): The width of transformed output.
        spatial_scale (float): Spatial scale factor to scale ROI coords. Default: 1.0

    Returns:
        Variable: The output of ROIPerspectiveTransformOp which is a 4-D tensor with shape 
                  (num_rois, channels, transformed_h, transformed_w).

    Examples:
        .. code-block:: python

            out = fluid.layers.roi_perspective_transform(input, rois, 7, 7, 1.0)
    """
    helper = LayerHelper('roi_perspective_transform', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
1622
    out = helper.create_variable_for_type_inference(dtype)
W
whs 已提交
1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
    helper.append_op(
        type="roi_perspective_transform",
        inputs={"X": input,
                "ROIs": rois},
        outputs={"Out": out},
        attrs={
            "transformed_height": transformed_height,
            "transformed_width": transformed_width,
            "spatial_scale": spatial_scale
        })
    return out


1636 1637
def generate_proposal_labels(rpn_rois,
                             gt_classes,
1638
                             is_crowd,
1639
                             gt_boxes,
1640
                             im_info,
1641 1642 1643 1644 1645 1646
                             batch_size_per_im=256,
                             fg_fraction=0.25,
                             fg_thresh=0.25,
                             bg_thresh_hi=0.5,
                             bg_thresh_lo=0.0,
                             bbox_reg_weights=[0.1, 0.1, 0.2, 0.2],
1647 1648
                             class_nums=None,
                             use_random=True):
1649 1650
    """
    ** Generate proposal labels Faster-RCNN **
B
buxingyuan 已提交
1651
    This operator can be, for given the GenerateProposalOp output bounding boxes and groundtruth,
B
buxingyuan 已提交
1652
    to sample foreground boxes and background boxes, and compute loss target.
B
buxingyuan 已提交
1653 1654 1655

    RpnRois is the output boxes of RPN and was processed by generate_proposal_op, these boxes
    were combined with groundtruth boxes and sampled according to batch_size_per_im and fg_fraction,
B
buxingyuan 已提交
1656
    If an instance with a groundtruth overlap greater than fg_thresh, then it was considered as a foreground sample.
B
buxingyuan 已提交
1657 1658
    If an instance with a groundtruth overlap greater than bg_thresh_lo and lower than bg_thresh_hi,
    then it was considered as a background sample.
B
buxingyuan 已提交
1659
    After all foreground and background boxes are chosen (so called Rois),
B
buxingyuan 已提交
1660
    then we apply random sampling to make sure
B
buxingyuan 已提交
1661
    the number of foreground boxes is no more than batch_size_per_im * fg_fraction.
B
buxingyuan 已提交
1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680

    For each box in Rois, we assign the classification (class label) and regression targets (box label) to it.
    Finally BboxInsideWeights and BboxOutsideWeights are used to specify whether it would contribute to training loss.

    Args:
        rpn_rois(Variable): A 2-D LoDTensor with shape [N, 4]. N is the number of the GenerateProposalOp's output, each element is a bounding box with [xmin, ymin, xmax, ymax] format.
        gt_classes(Variable): A 2-D LoDTensor with shape [M, 1]. M is the number of groundtruth, each element is a class label of groundtruth.
        is_crowd(Variable): A 2-D LoDTensor with shape [M, 1]. M is the number of groundtruth, each element is a flag indicates whether a groundtruth is crowd.
        gt_boxes(Variable): A 2-D LoDTensor with shape [M, 4]. M is the number of groundtruth, each element is a bounding box with [xmin, ymin, xmax, ymax] format.
        im_info(Variable): A 2-D LoDTensor with shape [B, 3]. B is the number of input images, each element consists of im_height, im_width, im_scale.

        batch_size_per_im(int): Batch size of rois per images.
        fg_fraction(float): Foreground fraction in total batch_size_per_im.
        fg_thresh(float): Overlap threshold which is used to chose foreground sample.
        bg_thresh_hi(float): Overlap threshold upper bound which is used to chose background sample.
        bg_thresh_lo(float): Overlap threshold lower bound which is used to chose background sample.
        bbox_reg_weights(list|tuple): Box regression weights.
        class_nums(int): Class number.
        use_random(bool): Use random sampling to choose foreground and background boxes.
1681 1682 1683 1684
    """

    helper = LayerHelper('generate_proposal_labels', **locals())

X
Xin Pan 已提交
1685 1686 1687 1688 1689 1690 1691 1692 1693
    rois = helper.create_variable_for_type_inference(dtype=rpn_rois.dtype)
    labels_int32 = helper.create_variable_for_type_inference(
        dtype=gt_classes.dtype)
    bbox_targets = helper.create_variable_for_type_inference(
        dtype=rpn_rois.dtype)
    bbox_inside_weights = helper.create_variable_for_type_inference(
        dtype=rpn_rois.dtype)
    bbox_outside_weights = helper.create_variable_for_type_inference(
        dtype=rpn_rois.dtype)
1694 1695 1696 1697 1698 1699

    helper.append_op(
        type="generate_proposal_labels",
        inputs={
            'RpnRois': rpn_rois,
            'GtClasses': gt_classes,
1700
            'IsCrowd': is_crowd,
1701
            'GtBoxes': gt_boxes,
1702
            'ImInfo': im_info
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
        },
        outputs={
            'Rois': rois,
            'LabelsInt32': labels_int32,
            'BboxTargets': bbox_targets,
            'BboxInsideWeights': bbox_inside_weights,
            'BboxOutsideWeights': bbox_outside_weights
        },
        attrs={
            'batch_size_per_im': batch_size_per_im,
            'fg_fraction': fg_fraction,
            'fg_thresh': fg_thresh,
            'bg_thresh_hi': bg_thresh_hi,
            'bg_thresh_lo': bg_thresh_lo,
            'bbox_reg_weights': bbox_reg_weights,
1718 1719
            'class_nums': class_nums,
            'use_random': use_random
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
        })

    rois.stop_gradient = True
    labels_int32.stop_gradient = True
    bbox_targets.stop_gradient = True
    bbox_inside_weights.stop_gradient = True
    bbox_outside_weights.stop_gradient = True

    return rois, labels_int32, bbox_targets, bbox_inside_weights, bbox_outside_weights


1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
def generate_proposals(scores,
                       bbox_deltas,
                       im_info,
                       anchors,
                       variances,
                       pre_nms_top_n=6000,
                       post_nms_top_n=1000,
                       nms_thresh=0.5,
                       min_size=0.1,
                       eta=1.0,
                       name=None):
    """
H
haowang101779990 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771
    **Generate proposal Faster-RCNN**

    This operation proposes RoIs according to each box with their probability to be a foreground object and 
    the box can be calculated by anchors. Bbox_deltais and scores to be an object are the output of RPN. Final proposals
    could be used to train detection net.

    For generating proposals, this operation performs following steps:

    1. Transposes and resizes scores and bbox_deltas in size of (H*W*A, 1) and (H*W*A, 4)
    2. Calculate box locations as proposals candidates. 
    3. Clip boxes to image
    4. Remove predicted boxes with small area. 
    5. Apply NMS to get final proposals as output.

    Args:
        scores(Variable): A 4-D Tensor with shape [N, A, H, W] represents the probability for each box to be an object.
            N is batch size, A is number of anchors, H and W are height and width of the feature map.
        bbox_deltas(Variable): A 4-D Tensor with shape [N, 4*A, H, W] represents the differece between predicted box locatoin and anchor location. 
        im_info(Variable): A 2-D Tensor with shape [N, 3] represents origin image information for N batch. Info contains height, width and scale
            between origin image size and the size of feature map.
        anchors(Variable):   A 4-D Tensor represents the anchors with a layout of [H, W, A, 4]. H and W are height and width of the feature map,
                    num_anchors is the box count of each position. Each anchor is in (xmin, ymin, xmax, ymax) format an unnormalized.
        variances(Variable): The expanded variances of anchors with a layout of [H, W, num_priors, 4]. Each variance is in (xcenter, ycenter, w, h) format.
        pre_nms_top_n(float): Number of total bboxes to be kept per image before NMS. 6000 by default.
        post_nms_top_n(float): Number of total bboxes to be kept per image after NMS. 1000 by default.
        nms_thresh(float): Threshold in NMS, 0.5 by default.
        min_size(float): Remove predicted boxes with either height or width < min_size. 0.1 by default.
        eta(float): Apply in adaptive NMS, if adaptive threshold > 0.5, adaptive_threshold = adaptive_threshold * eta in each iteration.

1772 1773 1774
    """
    helper = LayerHelper('generate_proposals', **locals())

X
Xin Pan 已提交
1775 1776 1777 1778
    rpn_rois = helper.create_variable_for_type_inference(
        dtype=bbox_deltas.dtype)
    rpn_roi_probs = helper.create_variable_for_type_inference(
        dtype=scores.dtype)
1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800
    helper.append_op(
        type="generate_proposals",
        inputs={
            'Scores': scores,
            'BboxDeltas': bbox_deltas,
            'ImInfo': im_info,
            'Anchors': anchors,
            'Variances': variances
        },
        attrs={
            'pre_nms_topN': pre_nms_top_n,
            'post_nms_topN': post_nms_top_n,
            'nms_thresh': nms_thresh,
            'min_size': min_size,
            'eta': eta
        },
        outputs={'RpnRois': rpn_rois,
                 'RpnRoiProbs': rpn_roi_probs})
    rpn_rois.stop_gradient = True
    rpn_roi_probs.stop_gradient = True

    return rpn_rois, rpn_roi_probs
J
jerrywgz 已提交
1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835


def box_clip(input_box, im_info, inplace=False, name=None):
    """
    Clip the box into the size given by im_info

    Args:
        input_box(variable): The input box, the last dimension is 4.
        im_info(variable): The information of image with shape [N, 3].
        inplace(bool): Must use :attr:`False` if :attr:`input_box` is used in 
                       multiple operators. If this flag is set :attr:`True`, 
                       reuse input :attr:`input_box` to clip, which will 
                       change the value of tensor variable :attr:`input_box` 
                       and might cause errors when :attr:`input_box` is used 
                       in multiple operators. If :attr:`False`, preserve the 
                       value pf :attr:`input_box` and create a new output 
                       tensor variable whose data is copied from input x but 
                       cliped.
        name (str): The name of this layer. It is optional.
    
    Returns:
        Variable: The cliped tensor variable.

    Examples:
        .. code-block:: python

            boxes = fluid.layers.data(
                name='data', shape=[8, 4], dtype='float32', lod_level=1)
            im_info = fluid.layers.data(name='im_info', shape=[3])
            out = fluid.layers.box_clip(
                input_box=boxes, im_info=im_info, inplace=True)
    """

    helper = LayerHelper("box_clip", **locals())
    output = helper.create_variable_for_type_inference(dtype=input_box.dtype)
J
jerrywgz 已提交
1836
    inputs = {"InputBox": input_box, "ImInfo": im_info}
J
jerrywgz 已提交
1837 1838 1839 1840 1841 1842 1843
    helper.append_op(
        type="box_clip",
        inputs=inputs,
        attrs={"inplace:": inplace},
        outputs={"OutputBox": output})

    return output