-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun3.py
270 lines (241 loc) · 10.9 KB
/
run3.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
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
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
from segment_anything import build_sam, SamPredictor
import os
import cv2
import numpy as np
from collections import defaultdict
import time
from classification.predict import Predict
class Colors:
# Ultralytics color palette https://ultralytics.com/
def __init__(self):
# hex = matplotlib.colors.TABLEAU_COLORS.values()
hexs = ('FF3838', 'FF9D97', 'FF701F', 'FFB21D', 'CFD231', '48F90A', '92CC17', '3DDB86', '1A9334', '00D4BB',
'2C99A8', '00C2FF', '344593', '6473FF', '0018EC', '8438FF', '520085', 'CB38FF', 'FF95C8', 'FF37C7')
self.palette = [self.hex2rgb(f'#{c}') for c in hexs]
self.n = len(self.palette)
def __call__(self, i, bgr=False):
c = self.palette[int(i) % self.n]
return (c[2], c[1], c[0]) if bgr else c
@staticmethod
def hex2rgb(h): # rgb order (PIL)
return tuple(int(h[1 + i:1 + i + 2], 16) for i in (0, 2, 4))
colors = Colors() # create instance for 'from utils.plots import colors'
from segment_anything import sam_model_registry, SamAutomaticMaskGenerator, SamPredictor
# predictor = SamPredictor(build_sam(checkpoint="checkpoint/sam_vit_b_01ec64.pth"))
# _ = predictor.model.to(device='cuda')
sam_checkpoint = "checkpoint/sam_vit_b_01ec64.pth"
model_type = "vit_b"
device = "cuda"
sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
predictor = SamPredictor(sam)
sam.to(device=device)
# image = cv2.imread('/home/hadoop-vacv/yanfeng/data/dancetrack/train/dancetrack0001/img1/00000109.jpg')
# image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# predictor.set_image(image)
# bbox = np.array([0,0,100,100], dtype=np.int32)
# masks, _, _ = predictor.predict(box=bbox)
# masks
# input_path = './img9'
# targets = [f for f in os.listdir(os.path.join(input_path, 'img1')) if
# not os.path.isdir(os.path.join(input_path, 'img1', f))]
# targets = [os.path.join(input_path, 'img1', f) for f in targets]
# targets.sort()
bboxes_all = defaultdict(list)
#gt_path = os.path.join(input_path, 'gt', 'gt.txt')
gt_path = "./yolo-pyqt/car_output.txt"
#print(gt_path)
#print(gt_path)
# gt_path = os.path.join('/home/hadoop-vacv/yanfeng/project/MOTRv2/MOTRv3/exps/motrv2ch_uni5cost6g/run2/tracker0', 'dancetrack0004.txt')
for l in open(gt_path):
#print(l)
t, i, *xywh, mark, label = l.strip().split(',')[:8]
print(t,i,mark,label)
t, i, mark, label = map(int, (t, i, mark, label))
if mark == 1:
continue
if label in [3, 4, 5, 6, 9, 10, 11]: # Non-person
continue
else:
crowd = False
x, y, w, h = map(int, map(float, (xywh)))
print(x,y,w,h)
bboxes_all[t].append([int(x), int(y),int(x+w),int(y+h), i])
#bboxes_all[t].append([int((x/640)*1920), int((y/640)*1080), int((w/640)*1920), int((h/640)*1080), i])
fps = 25
size = (3840, 2160)
videowriter = cv2.VideoWriter('tmp.avi', cv2.VideoWriter_fourcc('M', 'J', 'P', 'G'), fps, size)
video_path = "./yolo-pyqt/img11.mp4"
video = cv2.VideoCapture(video_path)
fps = int(video.get(cv2.CAP_PROP_FPS))
width = int(video.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(video.get(cv2.CAP_PROP_FRAME_HEIGHT))
print(width,height)
output_video = 'pig_video.mp4'
video_writer = cv2.VideoWriter(output_video, cv2.VideoWriter_fourcc(*'mp4v'), fps, (width, height))
frame_counter = 0
while True:
# 读取下一帧
ret, frame = video.read()
# 如果读取成功，则处理帧
if ret:
# 帧计数器增加
#print(frame_counter)
frame_counter += 1
image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
#cv2.imshow('ss',frame)
masks_all = []
#print(os.path.basename(t)[:-4])
bboxes = np.array(bboxes_all[frame_counter])
predictor.set_image(frame)
output_path = './img10/img10/'
for bbox in bboxes:
print(bbox)
masks, iou_predictions, low_res_masks = predictor.predict(box=bbox[:4])
index_max = iou_predictions.argsort()[0]
masks = np.concatenate(
[masks[index_max:(index_max + 1)], masks[index_max:(index_max + 1)], masks[index_max:(index_max + 1)]],
axis=0)
masks1 = masks
if masks1.dtype != np.uint8:
masks1 = (masks1 * 255).astype(np.uint8) # 乘以 255 并转换为 uint8
masks1 = 255 - masks1
# 保存掩膜图像
mask_image = masks1[index_max] # 假设我们只对单一的掩膜图像感兴趣
mask_image1 = mask_image[int(bbox[1]):int(bbox[3]), int(bbox[0]):int(bbox[2])]
# print(mask_image1.shape)
# print('frame',frame)
folder_path = f'{output_path}{bbox[-1]}-{int(int(frame_counter) / 100) + 1}/'
print('mask目录：',folder_path)
os.makedirs(folder_path, exist_ok=True)
cv2.imwrite(f"{folder_path}{frame_counter}.jpg", mask_image1)
masks = masks.astype(np.int32) * np.array(colors(bbox[4]))[:, None, None]
# cv2.imwrite('{}.jpg'.format(bbox[-1]),masks)
masks_all.append(masks)
predictor.reset_image()
if len(masks_all):
masks_sum = masks_all[0].copy()
for m in masks_all[1:]:
masks_sum += m
else:
print("error")
img = image.copy()[..., ::-1]
masks_sum = np.zeros_like(img).transpose(2, 0, 1)
img = image.copy()[..., ::-1]
img = (img * 0.5 + (masks_sum.transpose(1, 2, 0) * 30) % 128).astype(np.uint8)
for bbox in bboxes:
folder_path = f'{output_path}{bbox[-1]}-{int(int(frame_counter) / 100) + 1}/'
os.makedirs(folder_path, exist_ok=True)
output_cls = Predict(f"{folder_path}/{frame_counter}.jpg")
output_cls = str(output_cls)
bbox = list(bbox)
bbox.append(output_cls)
bbox = np.array(bbox)
# print(bbox)
txt_output_path = f"{output_path}{bbox[-2]}-{int(int(frame_counter) / 100) + 1}.txt"
with open(txt_output_path, 'a') as file:
# 写入数字1和2，然后添加一个空格
# file.write(str(((xmin + width) / 2) / 1920))
file.write(str(((int(bbox[0]) + int(bbox[2])) / 2) / 3840))
file.write(',')
# file.write(str(((ymin + height) / 2) / 1080))
file.write(str(((int(bbox[1]) + int(bbox[3])) / 2) / 2160))
file.write('\n')
# 写入换行符，以便开始新的一行
cv2.rectangle(img, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (0, 0, 255), thickness=3)
i = int(bbox[-2])
text = 'id:' + str(i)
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(img, text, (int(bbox[0]), int(bbox[1])), font, 1, (0, 255, 0), 2)
# print(bbox[-1])
output_class = str(bbox[-1])
cv2.putText(img, output_class, (int(bbox[2]) - 20, int(bbox[1])), font, 1, (0, 255, 0), 2)
cv2.namedWindow('pic', cv2.WINDOW_NORMAL)
cv2.resizeWindow('pic', 1280, 720)
cv2.imshow('pic',img)
cv2.waitKey(20)
video_writer.write(img)
cv2.imwrite(f"{output_path}/{frame_counter}.jpg", img)
print('picture目录：', f"{output_path}/{frame_counter}.jpg")
video_writer.release()
# for t in targets:
# print(f"Processing '{t}'...")
# image = cv2.imread(t)
# frame1 = t.split('\\')[-1] # 这会得到 '113.jpg'
# frame = frame1.split('.')[0] # 这会得到 '113'
# if image is None:
# print(f"Could not load '{t}' as an image, skipping...")
# continue
# image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# masks_all = []
# print(os.path.basename(t)[:-4])
# bboxes = np.array(bboxes_all[int(os.path.basename(t)[:-4])])
# # predictor.set_image(image)
# # masks, _, _ = predictor.predict(box=bboxes[:, :4])
# predictor.set_image(image)
# output_path = './img11/img11/'
# for bbox in bboxes:
# #print(bbox)
# masks, iou_predictions, low_res_masks = predictor.predict(box=bbox[:4])
# index_max = iou_predictions.argsort()[0]
# masks = np.concatenate(
# [masks[index_max:(index_max + 1)], masks[index_max:(index_max + 1)], masks[index_max:(index_max + 1)]],
# axis=0)
# masks1 = masks
# if masks1.dtype != np.uint8:
# masks1 = (masks1*255).astype(np.uint8) # 乘以 255 并转换为 uint8
# masks1 = 255 - masks1
# # 保存掩膜图像
# mask_image = masks1[index_max] # 假设我们只对单一的掩膜图像感兴趣
# mask_image1 = mask_image[int(bbox[1]):int(bbox[3]),int(bbox[0]):int(bbox[2])]
# #print(mask_image1.shape)
# #print('frame',frame)
# folder_path = f'{output_path}{bbox[-1]}-{int(int(frame)/100)+1}/'
# os.makedirs(folder_path, exist_ok=True)
# cv2.imwrite(f"{folder_path}{frame}.jpg", mask_image1)
# masks = masks.astype(np.int32) * np.array(colors(bbox[4]))[:, None, None]
# #cv2.imwrite('{}.jpg'.format(bbox[-1]),masks)
# masks_all.append(masks)
# predictor.reset_image()
#
# if len(masks_all):
# masks_sum = masks_all[0].copy()
# for m in masks_all[1:]:
# masks_sum += m
# else:
# masks_sum = np.zeros_like(img).transpose(2, 0, 1)
#
# img = image.copy()[..., ::-1]
# img = (img * 0.5 + (masks_sum.transpose(1, 2, 0) * 30) % 128).astype(np.uint8)
# for bbox in bboxes:
# folder_path = f'{output_path}{bbox[-1]}-{int(int(frame) / 100) + 1}/'
# os.makedirs(folder_path, exist_ok=True)
# output_cls = Predict(f"{folder_path}{frame}.jpg")
# output_cls = str(output_cls)
# bbox = list(bbox)
# bbox.append(output_cls)
# bbox = np.array(bbox)
# # print(bbox)
# txt_output_path = f"{output_path}{bbox[-2]}-{int(int(frame) / 100) + 1}.txt"
# with open(txt_output_path, 'a') as file:
# # 写入数字1和2，然后添加一个空格
# # file.write(str(((xmin + width) / 2) / 1920))
# file.write(str(((int(bbox[0]) + int(bbox[2])) / 2) / 2560))
# file.write(',')
# # file.write(str(((ymin + height) / 2) / 1080))
# file.write(str(((int(bbox[1]) + int(bbox[3])) / 2) / 1440))
# file.write('\n')
# # 写入换行符，以便开始新的一行
# cv2.rectangle(img, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (0, 0, 255), thickness=3)
# i = int(bbox[-2])
# text = 'id:' + str(i)
# font = cv2.FONT_HERSHEY_SIMPLEX
# cv2.putText(img, text, (int(bbox[0]), int(bbox[1])), font, 1, (0, 255, 0), 2)
# # print(bbox[-1])
# output_class = str(bbox[-1])
# cv2.putText(img, output_class, (int(bbox[2]) - 20, int(bbox[1])), font, 1, (0, 255, 0), 2)
#
# cv2.imwrite(f"{output_path}{frame}.jpg", img)
#
# videowriter.write(img)
#
# videowriter.release()