[cherry-pick] from_config hrnet, test=dygraph (#2219)

* from_config hrnet, test=dygraph

* update hrnet modelzoo, test=dygraph

dygraph/configs/hrnet/README.md

@@ -28,7 +28,7 @@
 
 ## Model Zoo
 
-| Backbone                | Type           | deformable Conv  | Image/gpu | Lr schd | Inf time (fps) | Box AP | Mask AP |                           Download                           | Configs |
-| :---------------------- | :------------- | :---: | :-------: | :-----: | :------------: | :----: | :-----: | :----------------------------------------------------------: | :-----: |
-| HRNetV2p_W18            | Faster         | False |     2     |   1x    |    -      |  35.7  |    -    | [model](https://paddlemodels.bj.bcebos.com/object_detection/dygraph/faster_rcnn_hrnetv2p_w18_1x_coco.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/dygraph/configs/hrnet/faster_rcnn_hrnetv2p_w18_1x_coco.yml) |
-| HRNetV2p_W18            | Faster         | False |     2     |   2x    |    -      |  37.7  |    -    | [model](https://paddlemodels.bj.bcebos.com/object_detection/dygraph/faster_rcnn_hrnetv2p_w18_2x_coco.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/dygraph/configs/hrnet/faster_rcnn_hrnetv2p_w18_2x_coco.yml) |
+| Backbone                | Type           | Image/gpu | Lr schd | Inf time (fps) | Box AP | Mask AP |                           Download                           | Configs |
+| :---------------------- | :------------- | :-------: | :-----: | :------------: | :----: | :-----: | :----------------------------------------------------------: | :-----: |
+| HRNetV2p_W18            | Faster         |     1     |   1x    |    -      |  36.8  |    -    | [model](https://paddlemodels.bj.bcebos.com/object_detection/dygraph/faster_rcnn_hrnetv2p_w18_1x_coco.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/dygraph/configs/hrnet/faster_rcnn_hrnetv2p_w18_1x_coco.yml) |
+| HRNetV2p_W18            | Faster         |     1     |   2x    |    -      |  39.0  |    -    | [model](https://paddlemodels.bj.bcebos.com/object_detection/dygraph/faster_rcnn_hrnetv2p_w18_2x_coco.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/dygraph/configs/hrnet/faster_rcnn_hrnetv2p_w18_2x_coco.yml) |

dygraph/configs/hrnet/_base_/faster_rcnn_hrnetv2p_w18.yml

 architecture: FasterRCNN
 pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/HRNet_W18_C_pretrained.tar
-weights: output/faster_rcnn_hrnetv2p_w18_1x_coco/model_final
 load_static_weights: True
 
-# Model Achitecture
 FasterRCNN:
-  # model anchor info flow
-  anchor: Anchor
-  proposal: Proposal
-  # model feat info flow
   backbone: HRNet
   neck: HRFPN
   rpn_head: RPNHead
@@ -26,65 +20,51 @@ HRFPN:
   share_conv: false
 
 RPNHead:
-  rpn_feat:
-    name: RPNFeat
-    feat_in: 256
-    feat_out: 256
-  anchor_per_position: 3
-  rpn_channel: 256
-
-Anchor:
   anchor_generator:
-    name: AnchorGeneratorRPN
     aspect_ratios: [0.5, 1.0, 2.0]
-    anchor_start_size: 32
-    stride: [4., 4.]
-  anchor_target_generator:
-    name: AnchorTargetGeneratorRPN
+    anchor_sizes: [[32], [64], [128], [256], [512]]
+    strides: [4, 8, 16, 32, 64]
+  rpn_target_assign:
     batch_size_per_im: 256
     fg_fraction: 0.5
     negative_overlap: 0.3
     positive_overlap: 0.7
-    straddle_thresh: 0.0
-
-Proposal:
-  proposal_generator:
-    name: ProposalGenerator
+    use_random: True
+  train_proposal:
+    min_size: 0.0
+    nms_thresh: 0.7
+    pre_nms_top_n: 2000
+    post_nms_top_n: 2000
+    topk_after_collect: True
+  test_proposal:
     min_size: 0.0
     nms_thresh: 0.7
-    train_pre_nms_top_n: 2000
-    train_post_nms_top_n: 2000
-    infer_pre_nms_top_n: 1000
-    infer_post_nms_top_n: 1000
-  proposal_target_generator:
-    name: ProposalTargetGenerator
-    batch_size_per_im: 512
-    bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
-    bg_thresh_hi: [0.5,]
-    bg_thresh_lo: [0.0,]
-    fg_thresh: [0.5,]
-    fg_fraction: 0.25
+    pre_nms_top_n: 1000
+    post_nms_top_n: 1000
 
 BBoxHead:
-  bbox_feat:
-    name: BBoxFeat
-    roi_extractor:
-      name: RoIAlign
-      resolution: 7
-      sampling_ratio: 2
-    head_feat:
-      name: TwoFCHead
-      in_dim: 256
-      mlp_dim: 1024
-  in_feat: 1024
+  head: TwoFCHead
+  roi_extractor:
+    resolution: 7
+    sampling_ratio: 0
+    aligned: True
+  bbox_assigner: BBoxAssigner
+
+BBoxAssigner:
+  batch_size_per_im: 512
+  bg_thresh: 0.5
+  fg_thresh: 0.5
+  fg_fraction: 0.25
+  use_random: True
+
+TwoFCHead:
+  mlp_dim: 1024
 
 BBoxPostProcess:
-  decode:
-    name: RCNNBox
-    num_classes: 81
-    batch_size: 1
+  decode: RCNNBox
   nms:
     name: MultiClassNMS
     keep_top_k: 100
     score_threshold: 0.05
     nms_threshold: 0.5
+    normalized: true

dygraph/configs/hrnet/faster_rcnn_hrnetv2p_w18_1x_coco.yml

@@ -6,6 +6,9 @@ _BASE_: [
   '../runtime.yml',
 ]
 
+weights: output/faster_rcnn_hrnetv2p_w18_1x_coco/model_final
+epoch: 12
+
 LearningRate:
   base_lr: 0.02
   schedulers:

dygraph/ppdet/modeling/backbones/hrnet.py

@@ -22,6 +22,7 @@ from numbers import Integral
 import math
 
 from ppdet.core.workspace import register, serializable
+from ..shape_spec import ShapeSpec
 
 __all__ = ['HRNet']
 
@@ -577,6 +578,8 @@ class HRNet(nn.Layer):
 
         channels_2, channels_3, channels_4 = self.channels[width]
         num_modules_2, num_modules_3, num_modules_4 = 1, 4, 3
+        self._out_channels = channels_4
+        self._out_strides = [4, 8, 16, 32]
 
         self.conv_layer1_1 = ConvNormLayer(
             ch_in=3,
@@ -666,3 +669,11 @@ class HRNet(nn.Layer):
                 res.append(layer)
 
         return res
+
+    @property
+    def out_shape(self):
+        return [
+            ShapeSpec(
+                channels=self._out_channels[i], stride=self._out_strides[i])
+            for i in self.return_idx
+        ]

dygraph/ppdet/modeling/necks/hrfpn.py

@@ -18,6 +18,7 @@ from paddle import ParamAttr
 import paddle.nn as nn
 from paddle.regularizer import L2Decay
 from ppdet.core.workspace import register, serializable
+from ..shape_spec import ShapeSpec
 
 __all__ = ['HRFPN']
 
@@ -26,23 +27,28 @@ __all__ = ['HRFPN']
 class HRFPN(nn.Layer):
     """
     Args:
-        in_channel (int): number of input feature channels from backbone
+        in_channels (list): number of input feature channels from backbone
         out_channel (int): number of output feature channels
         share_conv (bool): whether to share conv for different layers' reduction
-        spatial_scale (list): feature map scaling factor
+        spatial_scales (list): feature map scaling factor
+        extra_stage (int): add extra stage for returning HRFPN fpn_feats
     """
 
-    def __init__(
-            self,
-            in_channel=270,
-            out_channel=256,
-            share_conv=False,
-            spatial_scale=[1. / 4, 1. / 8, 1. / 16, 1. / 32, 1. / 64], ):
+    def __init__(self,
+                 in_channels=[18, 36, 72, 144],
+                 out_channel=256,
+                 share_conv=False,
+                 extra_stage=1,
+                 spatial_scales=[1. / 4, 1. / 8, 1. / 16, 1. / 32]):
         super(HRFPN, self).__init__()
+        in_channel = sum(in_channels)
         self.in_channel = in_channel
         self.out_channel = out_channel
         self.share_conv = share_conv
-        self.spatial_scale = spatial_scale
+        for i in range(extra_stage):
+            spatial_scales = spatial_scales + [spatial_scales[-1] / 2.]
+        self.spatial_scales = spatial_scales
+        self.num_out = len(self.spatial_scales)
 
         self.reduction = nn.Conv2D(
             in_channels=in_channel,
@@ -50,7 +56,7 @@ class HRFPN(nn.Layer):
             kernel_size=1,
             weight_attr=ParamAttr(name='hrfpn_reduction_weights'),
             bias_attr=False)
-        self.num_out = len(self.spatial_scale)
+
         if share_conv:
             self.fpn_conv = nn.Conv2D(
                 in_channels=out_channel,
@@ -106,5 +112,20 @@ class HRFPN(nn.Layer):
             conv = conv_func(outs[i])
             outputs.append(conv)
 
-        fpn_feat = [outputs[k] for k in range(self.num_out)]
-        return fpn_feat, self.spatial_scale
+        fpn_feats = [outputs[k] for k in range(self.num_out)]
+        return fpn_feats
+
+    @classmethod
+    def from_config(cls, cfg, input_shape):
+        return {
+            'in_channels': [i.channels for i in input_shape],
+            'spatial_scales': [1.0 / i.stride for i in input_shape],
+        }
+
+    @property
+    def out_shape(self):
+        return [
+            ShapeSpec(
+                channels=self.out_channel, stride=1. / s)
+            for s in self.spatial_scales
+        ]