-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain_posetrack.py
135 lines (118 loc) · 5.66 KB
/
main_posetrack.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
import os
import cv2
from det_seg_track.DetSegTrack import DetSegTrack
import time
import numpy as np
import pandas as pd
import argparse
import re
import json
from det_seg_track.utils import bbox_to_xywh, mask_to_bbox, getKeypointsVis, getPossibleFaceArea
#extract int numbers from string
def num_sort(test_string):
return list(map(int, re.findall(r'\d+', test_string)))[0]
parser = argparse.ArgumentParser(description='Project')
parser.add_argument('--detector', required=True,
help='detector')
parser.add_argument('--tracker', required=True,
help='tracker')
parser.add_argument('--segmentator', required=False, default=None,
help='segmentator')
parser.add_argument('--use_deployed_model',type=bool, required=False, default=False,
help='use_deployed_model')
parser.add_argument('--save_json', type=bool, default=False, required=False,
help='save results')
parser.add_argument('--save_vis', type=bool, default=False, required=False,
help='save results')
if __name__ == "__main__":
args = parser.parse_args()
detector = args.detector
tracker = args.tracker
segmentator = args.segmentator
use_deployed_model = args.use_deployed_model
save_json = args.save_json
save_vis = args.save_vis
else:
vid_name = 'IMG_9297'
#vid_path = os.path.join('/mnt','c','Users','Dell','studia-pliki-robocze','magisterka','src', vid_name + '.mov')
# python main.py --filepath /mnt/c/Users/Dell/studia-pliki-robocze/magisterka/src/IMG_4827.mp4 --detector rtmo-l --tracker bytetrack --segmentator mobile_sam
# python main.py --filepath /mnt/d/test/iphone/video/IMG_9297.mp4 --detector rtmo-l --tracker bytetrack --segmentator mobile_sam
vid_path = os.path.join('/mnt','d', 'test','iphone','video' , vid_name + '.mp4')
detector = 'rtmo-l'
tracker = 'bytetrack'
segmentator = 'mobile_sam'
use_deployed_model = False
save_mode = 'vid'
hr_estimation_method = None
out_dir = './results/keypoints'
#out_img_dir = './results/images'
out_img_dir = os.path.join('/mnt','d', 'test','PoseTrack21-main','preds')
input_path = os.path.join('/mnt','d', 'test','PoseTrack21-main','downloads')
gt_base_path = os.path.join(input_path, 'posetrack_data','val')
img_dirs = os.listdir(gt_base_path)
try:
for dir_name in img_dirs:
det_seg_track_module = DetSegTrack(detector, tracker, segmentator, use_deployed_model = use_deployed_model, validate_person= False)
# read gt data and init results
gt_path = os.path.join(gt_base_path, dir_name)
with open(gt_path, 'r') as f:
gt_data = json.load(f)
results = {
'images': gt_data['images'],
'annotations': [],
'categories': gt_data['categories']
}
# gt annotations to df
annotations_gt = pd.DataFrame.from_dict(gt_data['annotations'])
ids = []
for img_data in gt_data['images']:
if img_data['is_labeled']:
img_name = img_data['file_name']
print('Frame: ', img_name, '\n')
img_id = img_data['image_id']
# img_gt_df = annotations_gt[annotations_gt['image_id'] == img_id]
image_path = os.path.join(input_path, img_name)
frame = cv2.imread(image_path)
annotated_frame, person_list, params = det_seg_track_module.estimate(frame, visualize=save_vis)
for person in person_list:
# gt_person_df = img_gt_df[img_gt_df['track_id'] == person.tracker_id]
#if person.tracker_id is None:
# person.tracker_id = max(ids) + 100
#ids.append(person.tracker_id)
possible_face_mask = getPossibleFaceArea(person, frame.shape[:2])
if possible_face_mask is not None:
mask_bbox = bbox_to_xywh(mask_to_bbox(possible_face_mask))
else:
mask_bbox = [0,0,0,0]
person_dict = {
'bbox': bbox_to_xywh(person.bbox),
'bbox_head': mask_bbox,
'category_id': 1,
'id': int(str(img_id) + '0' + str(person.tracker_id)),
'image_id': int(img_id),
'keypoints': getKeypointsVis(person.keypoints, frame.shape),
'person_id': int(person.tracker_id),
'track_id': int(person.tracker_id)
}
results['annotations'].append(person_dict)
if save_vis:
# bbox =
#annotated_frame = cv2.rectangle(annotated_frame, (bbox[0], bbox[1]), (bbox[2], bbox[3]),
# (0, 255, 255), 5)
file_name = img_name.split("/")[-1]
folder = img_name.split("/")[-2]
out_path_img = os.path.join(out_img_dir, folder + '_' + file_name)
cv2.imwrite(out_path_img, annotated_frame)
else:
continue
if save_json:
out_dir_path = os.path.join(out_dir, dir_name)
with open(out_dir_path , 'w', encoding='utf-8') as f:
json.dump(results, f, ensure_ascii=False, indent=4)
del det_seg_track_module
except KeyboardInterrupt:
print('\nKeyboard interrupt \n')
if save_json:
out_dir_path = os.path.join(out_dir, dir_name)
with open(out_dir_path , 'w', encoding='utf-8') as f:
json.dump(results, f, ensure_ascii=False, indent=4)