# ------------------------------------------------------------------------ # Copyright (c) 2021 megvii-model. All Rights Reserved. # ------------------------------------------------------------------------ # Modified from Deformable DETR (https://github.com/fundamentalvision/Deformable-DETR) # Copyright (c) 2020 SenseTime. All Rights Reserved. # ------------------------------------------------------------------------ # Modified from DETR (https://github.com/facebookresearch/detr) # Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved # ------------------------------------------------------------------------ """ MOT dataset which returns image_id for evaluation. """ from pathlib import Path import cv2 import numpy as np import torch import torch.utils.data import os.path as osp from PIL import Image, ImageDraw import copy import datasets.transforms as T from models.structures import Instances class DetMOTDetection: def __init__(self, args, data_txt_path: str, seqs_folder, dataset2transform): self.args = args self.dataset2transform = dataset2transform self.num_frames_per_batch = max(args.sampler_lengths) self.sample_mode = args.sample_mode self.sample_interval = args.sample_interval self.vis = args.vis self.video_dict = {} with open(data_txt_path, 'r') as file: self.img_files = file.readlines() self.img_files = [osp.join(seqs_folder, x.strip()) for x in self.img_files] self.img_files = list(filter(lambda x: len(x) > 0, self.img_files)) self.label_files = [(x.replace('images', 'labels_with_ids').replace('.png', '.txt').replace('.jpg', '.txt')) for x in self.img_files] # The number of images per sample: 1 + (num_frames - 1) * interval. # The number of valid samples: num_images - num_image_per_sample + 1. self.item_num = len(self.img_files) - (self.num_frames_per_batch - 1) * self.sample_interval self._register_videos() # video sampler. self.sampler_steps: list = args.sampler_steps self.lengths: list = args.sampler_lengths print("sampler_steps={} lenghts={}".format(self.sampler_steps, self.lengths)) if self.sampler_steps is not None and len(self.sampler_steps) > 0: # Enable sampling length adjustment. assert len(self.lengths) > 0 assert len(self.lengths) == len(self.sampler_steps) + 1 for i in range(len(self.sampler_steps) - 1): assert self.sampler_steps[i] < self.sampler_steps[i + 1] self.item_num = len(self.img_files) - (self.lengths[-1] - 1) * self.sample_interval self.period_idx = 0 self.num_frames_per_batch = self.lengths[0] self.current_epoch = 0 def _register_videos(self): for label_name in self.label_files: video_name = '/'.join(label_name.split('/')[:-1]) if video_name not in self.video_dict: print("register {}-th video: {} ".format(len(self.video_dict) + 1, video_name)) self.video_dict[video_name] = len(self.video_dict) # assert len(self.video_dict) <= 300 def set_epoch(self, epoch): self.current_epoch = epoch if self.sampler_steps is None or len(self.sampler_steps) == 0: # fixed sampling length. return for i in range(len(self.sampler_steps)): if epoch >= self.sampler_steps[i]: self.period_idx = i + 1 print("set epoch: epoch {} period_idx={}".format(epoch, self.period_idx)) self.num_frames_per_batch = self.lengths[self.period_idx] def step_epoch(self): # one epoch finishes. print("Dataset: epoch {} finishes".format(self.current_epoch)) self.set_epoch(self.current_epoch + 1) @staticmethod def _targets_to_instances(targets: dict, img_shape) -> Instances: gt_instances = Instances(tuple(img_shape)) gt_instances.boxes = targets['boxes'] gt_instances.labels = targets['labels'] gt_instances.obj_ids = targets['obj_ids'] gt_instances.area = targets['area'] return gt_instances def _pre_single_frame(self, idx: int): img_path = self.img_files[idx] label_path = self.label_files[idx] if 'crowdhuman' in img_path: img_path = img_path.replace('.jpg', '.png') img = Image.open(img_path) targets = {} w, h = img._size assert w > 0 and h > 0, "invalid image {} with shape {} {}".format(img_path, w, h) if osp.isfile(label_path): labels0 = np.loadtxt(label_path, dtype=np.float32).reshape(-1, 6) # normalized cewh to pixel xyxy format labels = labels0.copy() labels[:, 2] = w * (labels0[:, 2] - labels0[:, 4] / 2) labels[:, 3] = h * (labels0[:, 3] - labels0[:, 5] / 2) labels[:, 4] = w * (labels0[:, 2] + labels0[:, 4] / 2) labels[:, 5] = h * (labels0[:, 3] + labels0[:, 5] / 2) else: raise ValueError('invalid label path: {}'.format(label_path)) video_name = '/'.join(label_path.split('/')[:-1]) obj_idx_offset = self.video_dict[video_name] * 1000000 # 1000000 unique ids is enough for a video. if 'crowdhuman' in img_path: targets['dataset'] = 'CrowdHuman' elif 'MOT17' in img_path: targets['dataset'] = 'MOT17' else: raise NotImplementedError() targets['boxes'] = [] targets['area'] = [] targets['iscrowd'] = [] targets['labels'] = [] targets['obj_ids'] = [] targets['image_id'] = torch.as_tensor(idx) targets['size'] = torch.as_tensor([h, w]) targets['orig_size'] = torch.as_tensor([h, w]) for label in labels: targets['boxes'].append(label[2:6].tolist()) targets['area'].append(label[4] * label[5]) targets['iscrowd'].append(0) targets['labels'].append(0) obj_id = label[1] + obj_idx_offset if label[1] >= 0 else label[1] targets['obj_ids'].append(obj_id) # relative id targets['area'] = torch.as_tensor(targets['area']) targets['iscrowd'] = torch.as_tensor(targets['iscrowd']) targets['labels'] = torch.as_tensor(targets['labels']) targets['obj_ids'] = torch.as_tensor(targets['obj_ids']) targets['boxes'] = torch.as_tensor(targets['boxes'], dtype=torch.float32).reshape(-1, 4) # targets['boxes'][:, 0::2].clamp_(min=0, max=w) # targets['boxes'][:, 1::2].clamp_(min=0, max=h) return img, targets def _get_sample_range(self, start_idx): # take default sampling method for normal dataset. assert self.sample_mode in ['fixed_interval', 'random_interval'], 'invalid sample mode: {}'.format(self.sample_mode) if self.sample_mode == 'fixed_interval': sample_interval = self.sample_interval elif self.sample_mode == 'random_interval': sample_interval = np.random.randint(1, self.sample_interval + 1) default_range = start_idx, start_idx + (self.num_frames_per_batch - 1) * sample_interval + 1, sample_interval return default_range def pre_continuous_frames(self, start, end, interval=1): targets = [] images = [] for i in range(start, end, interval): img_i, targets_i = self._pre_single_frame(i) images.append(img_i) targets.append(targets_i) return images, targets def __getitem__(self, idx): sample_start, sample_end, sample_interval = self._get_sample_range(idx) images, targets = self.pre_continuous_frames(sample_start, sample_end, sample_interval) data = {} dataset_name = targets[0]['dataset'] transform = self.dataset2transform[dataset_name] if transform is not None: images, targets = transform(images, targets) gt_instances = [] for img_i, targets_i in zip(images, targets): gt_instances_i = self._targets_to_instances(targets_i, img_i.shape[1:3]) gt_instances.append(gt_instances_i) data.update({ 'imgs': images, 'gt_instances': gt_instances, }) if self.args.vis: data['ori_img'] = [target_i['ori_img'] for target_i in targets] return data def __len__(self): return self.item_num class DetMOTDetectionValidation(DetMOTDetection): def __init__(self, args, seqs_folder, dataset2transform): args.data_txt_path = args.val_data_txt_path super().__init__(args, seqs_folder, dataset2transform) def make_transforms_for_mot17(image_set, args=None): normalize = T.MotCompose([ T.MotToTensor(), T.MotNormalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) ]) scales = [608, 640, 672, 704, 736, 768, 800, 832, 864, 896, 928, 960, 992] if image_set == 'train': return T.MotCompose([ T.MotRandomHorizontalFlip(), T.MotRandomSelect( T.MotRandomResize(scales, max_size=1536), T.MotCompose([ T.MotRandomResize([400, 500, 600]), T.FixedMotRandomCrop(384, 600), T.MotRandomResize(scales, max_size=1536), ]) ), normalize, ]) if image_set == 'val': return T.MotCompose([ T.MotRandomResize([800], max_size=1333), normalize, ]) raise ValueError(f'unknown {image_set}') def make_transforms_for_crowdhuman(image_set, args=None): normalize = T.MotCompose([ T.MotToTensor(), T.MotNormalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) ]) scales = [608, 640, 672, 704, 736, 768, 800, 832, 864, 896, 928, 960, 992] if image_set == 'train': return T.MotCompose([ T.MotRandomHorizontalFlip(), T.FixedMotRandomShift(bs=1), T.MotRandomSelect( T.MotRandomResize(scales, max_size=1536), T.MotCompose([ T.MotRandomResize([400, 500, 600]), T.FixedMotRandomCrop(384, 600), T.MotRandomResize(scales, max_size=1536), ]) ), normalize, ]) if image_set == 'val': return T.MotCompose([ T.MotRandomResize([800], max_size=1333), normalize, ]) raise ValueError(f'unknown {image_set}') def build_dataset2transform(args, image_set): mot17_train = make_transforms_for_mot17('train', args) mot17_test = make_transforms_for_mot17('val', args) crowdhuman_train = make_transforms_for_crowdhuman('train', args) dataset2transform_train = {'MOT17': mot17_train, 'CrowdHuman': crowdhuman_train} dataset2transform_val = {'MOT17': mot17_test, 'CrowdHuman': mot17_test} if image_set == 'train': return dataset2transform_train elif image_set == 'val': return dataset2transform_val else: raise NotImplementedError() def build(image_set, args): root = Path(args.mot_path) assert root.exists(), f'provided MOT path {root} does not exist' dataset2transform = build_dataset2transform(args, image_set) if image_set == 'train': data_txt_path = args.data_txt_path_train dataset = DetMOTDetection(args, data_txt_path=data_txt_path, seqs_folder=root, dataset2transform=dataset2transform) if image_set == 'val': data_txt_path = args.data_txt_path_val dataset = DetMOTDetection(args, data_txt_path=data_txt_path, seqs_folder=root, dataset2transform=dataset2transform) return dataset