first_order_predictor.py 13.7 KB
Newer Older
L
LielinJiang 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
#  Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
#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
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
#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.

import os
import sys
17 18
import cv2
import math
L
LielinJiang 已提交
19 20 21 22 23 24 25 26 27

import yaml
import pickle
import imageio
import numpy as np
from tqdm import tqdm
from scipy.spatial import ConvexHull

import paddle
L
LielinJiang 已提交
28
from ppgan.utils.download import get_path_from_url
L
LielinJiang 已提交
29 30 31
from ppgan.utils.animate import normalize_kp
from ppgan.modules.keypoint_detector import KPDetector
from ppgan.models.generators.occlusion_aware import OcclusionAwareGenerator
32
from ppgan.faceutils import face_detection
L
LielinJiang 已提交
33 34 35

from .base_predictor import BasePredictor

36
IMAGE_SIZE = 256
L
LielinJiang 已提交
37 38 39 40 41 42 43 44 45

class FirstOrderPredictor(BasePredictor):
    def __init__(self,
                 output='output',
                 weight_path=None,
                 config=None,
                 relative=False,
                 adapt_scale=False,
                 find_best_frame=False,
46
                 best_frame=None,
47
                 ratio=1.0,
L
lijianshe02 已提交
48
                 filename='result.mp4',
F
FNRE 已提交
49 50
                 face_detector='sfd',
                 multi_person=False):
L
LielinJiang 已提交
51
        if config is not None and isinstance(config, str):
F
FNRE 已提交
52 53
            with open(config) as f:
                self.cfg = yaml.load(f, Loader=yaml.SafeLoader)
L
LielinJiang 已提交
54 55 56 57
        elif isinstance(config, dict):
            self.cfg = config
        elif config is None:
            self.cfg = {
F
FNRE 已提交
58
                'model': {
L
LielinJiang 已提交
59 60 61 62 63
                    'common_params': {
                        'num_kp': 10,
                        'num_channels': 3,
                        'estimate_jacobian': True
                    },
F
FNRE 已提交
64 65 66 67
                    'generator': {
                        'kp_detector_cfg': {
                            'temperature': 0.1,
                            'block_expansion': 32,
L
LielinJiang 已提交
68
                            'max_features': 1024,
F
FNRE 已提交
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
                            'scale_factor': 0.25,
                            'num_blocks': 5
                        },
                        'generator_cfg': {
                            'block_expansion': 64,
                            'max_features': 512,
                            'num_down_blocks': 2,
                            'num_bottleneck_blocks': 6,
                            'estimate_occlusion_map': True,
                            'dense_motion_params': {
                                'block_expansion': 64,
                                'max_features': 1024,
                                'num_blocks': 5,
                                'scale_factor': 0.25
                            }
L
LielinJiang 已提交
84 85 86 87 88 89
                        }
                    }
                }
            }
            if weight_path is None:
                vox_cpk_weight_url = 'https://paddlegan.bj.bcebos.com/applications/first_order_model/vox-cpk.pdparams'
L
LielinJiang 已提交
90
                weight_path = get_path_from_url(vox_cpk_weight_url)
L
LielinJiang 已提交
91 92

        self.weight_path = weight_path
93 94
        if not os.path.exists(output):
            os.makedirs(output)
L
LielinJiang 已提交
95
        self.output = output
96
        self.filename = filename
L
LielinJiang 已提交
97 98 99 100
        self.relative = relative
        self.adapt_scale = adapt_scale
        self.find_best_frame = find_best_frame
        self.best_frame = best_frame
101
        self.ratio = ratio
L
lijianshe02 已提交
102
        self.face_detector = face_detector
L
LielinJiang 已提交
103 104
        self.generator, self.kp_detector = self.load_checkpoints(
            self.cfg, self.weight_path)
F
FNRE 已提交
105
        self.multi_person = multi_person
L
LielinJiang 已提交
106

F
FNRE 已提交
107 108 109 110 111 112 113 114 115
    def read_img(self, path):
        img = imageio.imread(path)
        if img.ndim == 2:
            img = np.expand_dims(img, axis=2)
        # som images have 4 channels
        if img.shape[2] > 3:
            img = img[:,:,:3]
        return img

L
LielinJiang 已提交
116
    def run(self, source_image, driving_video):
F
FNRE 已提交
117
        def get_prediction(face_image):
118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
            if self.find_best_frame or self.best_frame is not None:
                i = self.best_frame if self.best_frame is not None else self.find_best_frame_func(
                    source_image, driving_video)

                print("Best frame: " + str(i))
                driving_forward = driving_video[i:]
                driving_backward = driving_video[:(i + 1)][::-1]
                predictions_forward = self.make_animation(
                    face_image,
                    driving_forward,
                    self.generator,
                    self.kp_detector,
                    relative=self.relative,
                    adapt_movement_scale=self.adapt_scale)
                predictions_backward = self.make_animation(
                    face_image,
                    driving_backward,
                    self.generator,
                    self.kp_detector,
                    relative=self.relative,
                    adapt_movement_scale=self.adapt_scale)
                predictions = predictions_backward[::-1] + predictions_forward[
                    1:]
            else:
                predictions = self.make_animation(
                    face_image,
                    driving_video,
                    self.generator,
                    self.kp_detector,
                    relative=self.relative,
                    adapt_movement_scale=self.adapt_scale)
F
FNRE 已提交
149
            return predictions
150

F
FNRE 已提交
151
        source_image = self.read_img(source_image)
F
FNRE 已提交
152 153 154 155 156 157 158
        reader = imageio.get_reader(driving_video)
        fps = reader.get_meta_data()['fps']
        driving_video = []
        try:
            for im in reader:
                driving_video.append(im)
        except RuntimeError:
F
FNRE 已提交
159
            print("Read driving video error!")
F
FNRE 已提交
160 161 162 163
            pass
        reader.close()

        driving_video = [
164
            cv2.resize(frame, (IMAGE_SIZE, IMAGE_SIZE)) / 255.0 for frame in driving_video
F
FNRE 已提交
165 166 167 168 169 170
        ]
        results = []

        # for single person
        if not self.multi_person:
            h, w, _ = source_image.shape
171
            source_image = cv2.resize(source_image, (IMAGE_SIZE, IMAGE_SIZE)) / 255.0
F
FNRE 已提交
172 173 174 175
            predictions = get_prediction(source_image)
            imageio.mimsave(os.path.join(self.output, self.filename), [
                cv2.resize((frame * 255.0).astype('uint8'), (h, w))
                for frame in predictions
F
FNRE 已提交
176 177
            ],
                            fps=fps)
F
FNRE 已提交
178 179 180 181 182 183
            return

        bboxes = self.extract_bbox(source_image.copy())
        print(str(len(bboxes)) + " persons have been detected")
        if len(bboxes) <= 1:
            h, w, _ = source_image.shape
184
            source_image = cv2.resize(source_image, (IMAGE_SIZE, IMAGE_SIZE)) / 255.0
F
FNRE 已提交
185 186 187 188
            predictions = get_prediction(source_image)
            imageio.mimsave(os.path.join(self.output, self.filename), [
                cv2.resize((frame * 255.0).astype('uint8'), (h, w))
                for frame in predictions
F
FNRE 已提交
189 190
            ],
                            fps=fps)
F
FNRE 已提交
191 192 193 194 195
            return

        # for multi person
        for rec in bboxes:
            face_image = source_image.copy()[rec[1]:rec[3], rec[0]:rec[2]]
196
            face_image = cv2.resize(face_image, (IMAGE_SIZE, IMAGE_SIZE)) / 255.0
F
FNRE 已提交
197
            predictions = get_prediction(face_image)
198 199 200
            results.append({'rec': rec, 'predict': predictions})

        out_frame = []
201

202 203 204
        for i in range(len(driving_video)):
            frame = source_image.copy()
            for result in results:
F
FNRE 已提交
205
                x1, y1, x2, y2, _ = result['rec']
206 207 208 209
                h = y2 - y1
                w = x2 - x1
                out = result['predict'][i] * 255.0
                out = cv2.resize(out.astype(np.uint8), (x2 - x1, y2 - y1))
210 211 212 213 214 215 216 217 218 219 220
                if len(results) == 1:
                    frame[y1:y2, x1:x2] = out
                else:
                    patch = np.zeros(frame.shape).astype('uint8')
                    patch[y1:y2, x1:x2] = out
                    mask = np.zeros(frame.shape[:2]).astype('uint8')
                    cx = int((x1 + x2) / 2)
                    cy = int((y1 + y2) / 2)
                    cv2.circle(mask, (cx, cy), math.ceil(h * self.ratio),
                               (255, 255, 255), -1, 8, 0)
                    frame = cv2.copyTo(patch, mask, frame)
221 222

            out_frame.append(frame)
223
        imageio.mimsave(os.path.join(self.output, self.filename),
224 225
                        [frame for frame in out_frame],
                        fps=fps)
L
LielinJiang 已提交
226 227 228 229

    def load_checkpoints(self, config, checkpoint_path):

        generator = OcclusionAwareGenerator(
F
FNRE 已提交
230 231
            **config['model']['generator']['generator_cfg'],
            **config['model']['common_params'])
L
LielinJiang 已提交
232

F
FNRE 已提交
233 234 235
        kp_detector = KPDetector(
            **config['model']['generator']['kp_detector_cfg'],
            **config['model']['common_params'])
L
LielinJiang 已提交
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305

        checkpoint = paddle.load(self.weight_path)
        generator.set_state_dict(checkpoint['generator'])

        kp_detector.set_state_dict(checkpoint['kp_detector'])

        generator.eval()
        kp_detector.eval()

        return generator, kp_detector

    def make_animation(self,
                       source_image,
                       driving_video,
                       generator,
                       kp_detector,
                       relative=True,
                       adapt_movement_scale=True):
        with paddle.no_grad():
            predictions = []
            source = paddle.to_tensor(source_image[np.newaxis].astype(
                np.float32)).transpose([0, 3, 1, 2])

            driving = paddle.to_tensor(
                np.array(driving_video)[np.newaxis].astype(
                    np.float32)).transpose([0, 4, 1, 2, 3])
            kp_source = kp_detector(source)
            kp_driving_initial = kp_detector(driving[:, :, 0])

            for frame_idx in tqdm(range(driving.shape[2])):
                driving_frame = driving[:, :, frame_idx]
                kp_driving = kp_detector(driving_frame)
                kp_norm = normalize_kp(
                    kp_source=kp_source,
                    kp_driving=kp_driving,
                    kp_driving_initial=kp_driving_initial,
                    use_relative_movement=relative,
                    use_relative_jacobian=relative,
                    adapt_movement_scale=adapt_movement_scale)
                out = generator(source, kp_source=kp_source, kp_driving=kp_norm)

                predictions.append(
                    np.transpose(out['prediction'].numpy(), [0, 2, 3, 1])[0])
        return predictions

    def find_best_frame_func(self, source, driving):
        import face_alignment

        def normalize_kp(kp):
            kp = kp - kp.mean(axis=0, keepdims=True)
            area = ConvexHull(kp[:, :2]).volume
            area = np.sqrt(area)
            kp[:, :2] = kp[:, :2] / area
            return kp

        fa = face_alignment.FaceAlignment(face_alignment.LandmarksType._2D,
                                          flip_input=True)

        kp_source = fa.get_landmarks(255 * source)[0]
        kp_source = normalize_kp(kp_source)
        norm = float('inf')
        frame_num = 0
        for i, image in tqdm(enumerate(driving)):
            kp_driving = fa.get_landmarks(255 * image)[0]
            kp_driving = normalize_kp(kp_driving)
            new_norm = (np.abs(kp_source - kp_driving)**2).sum()
            if new_norm < norm:
                norm = new_norm
                frame_num = i
        return frame_num
306 307 308

    def extract_bbox(self, image):
        detector = face_detection.FaceAlignment(
L
lijianshe02 已提交
309 310 311
            face_detection.LandmarksType._2D,
            flip_input=False,
            face_detector=self.face_detector)
312 313 314

        frame = [image]
        predictions = detector.get_detections_for_image(np.array(frame))
F
FNRE 已提交
315 316 317
        person_num = len(predictions)
        if person_num == 0:
            return np.array([])
318
        results = []
F
FNRE 已提交
319
        face_boxs = []
320 321 322 323 324 325 326 327
        h, w, _ = image.shape
        for rect in predictions:
            bh = rect[3] - rect[1]
            bw = rect[2] - rect[0]
            cy = rect[1] + int(bh / 2)
            cx = rect[0] + int(bw / 2)
            margin = max(bh, bw)
            y1 = max(0, cy - margin)
328
            x1 = max(0, cx - int(0.8 * margin))
329
            y2 = min(h, cy + margin)
330
            x2 = min(w, cx + int(0.8 * margin))
F
FNRE 已提交
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
            area = (y2 - y1) * (x2 - x1)
            results.append([x1, y1, x2, y2, area])
        # if a person has more than one bbox, keep the largest one
        # maybe greedy will be better?
        sorted(results, key=lambda area: area[4], reverse=True)
        results_box = [results[0]]
        for i in range(1, person_num):
            num = len(results_box)
            add_person = True
            for j in range(num):
                pre_person = results_box[j]
                iou = self.IOU(pre_person[0], pre_person[1], pre_person[2],
                               pre_person[3], pre_person[4], results[i][0],
                               results[i][1], results[i][2], results[i][3],
                               results[i][4])
                if iou > 0.5:
                    add_person = False
                    break
            if add_person:
                results_box.append(results[i])
        boxes = np.array(results_box)
352
        return boxes
F
FNRE 已提交
353 354 355 356 357 358 359 360 361 362 363 364

    def IOU(self, ax1, ay1, ax2, ay2, sa, bx1, by1, bx2, by2, sb):
        #sa = abs((ax2 - ax1) * (ay2 - ay1))
        #sb = abs((bx2 - bx1) * (by2 - by1))
        x1, y1 = max(ax1, bx1), max(ay1, by1)
        x2, y2 = min(ax2, bx2), min(ay2, by2)
        w = x2 - x1
        h = y2 - y1
        if w < 0 or h < 0:
            return 0.0
        else:
            return 1.0 * w * h / (sa + sb - w * h)