-
Notifications
You must be signed in to change notification settings - Fork 3
/
sequence.py
285 lines (249 loc) · 11.4 KB
/
sequence.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
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
"""
Class to represent a Video Seqeunce
"""
from annotations import Annotations
import numpy as np
from PIL import Image
from PIL import ImageDraw
import os
import cv2
from tqdm import tqdm
COLOR = [(255, 0, 0),
(0 , 162, 232),
(136, 0 , 2255),
(128, 128, 128),
(0, 255, 0),
(255, 0, 255),
(255, 255, 0),
(0, 0, 255),
(0, 255, 255),
(255, 127, 39),
(0.2,0.7,0.255),
(0.7,0.2,0.255),
(65, 51, 178),
(0, 0, 0)]
class Sequence:
def __init__(self, image_dir, gt_path, seqname=None, trackers=[]):
if (seqname == None):
# Assumed data format is /path/to/data/seqname.{csv|txt}
self.seqname = gt_path.split('/')[-1][:-4]
else:
self.seqname = seqname
self.gt_annotations = Annotations(gt_path, seqname=seqname)
self.tracker_res = {}
for i in trackers:
try:
self.tracker_res[i.name] = i.get_res_of(seqname)
except:
print(self.seqname, 'not available for', i.name)
self.img_dir = image_dir
self.images = [i for i in os.listdir(image_dir) if i[-4:] == '.png' or i[-4:] == '.jpg']
self.images.sort()
height, width, layers = cv2.imread(os.path.join(image_dir, self.images[0])).shape
self.height = height
self.width = width
self.size = (width, height)
self.obj_size = self.gt_annotations.obj_size
def get_frame(self, frame_num, scale_bb=1, sort_order='x',
off_x=0, off_y=0, show_boxes=True, boxColor=0,
show_text=True, textColor=0, save_path=None):
img_path = os.path.join(self.img_dir, self.images[frame_num])
try:
with Image.open(img_path) as image:
#print("Mode", image.mode, type(image.mode))
if (image.mode != 'L'):
image = image.convert(mode='L')
draw = ImageDraw.Draw(image, 'L')
dets = self.gt_annotations.get_frame(frame_num+1)[:,1:].astype(np.float)
if len(dets) == 0:
if (save_path is not None):
image.save(save_path)
return image
# Format: [frame, ID, x1, y1, width, height, obj_class, species, occluded, noisy_frame]
labels = None
#print('sort_order', sort_order)
if (sort_order == 'x+y'):
d = dets[np.argsort(dets[:,2] + dets[:,1])]
else:
d = dets[np.lexsort((dets[:,2],dets[:,1]))]
ids = d[:,0]
#print(ids)
h,w = image.size
#frame_text = "Frame: " + frame_num + "ids on screen: \n" + str(ids) + "\n count = " + str(len(ids))
frame_text = "Frame: {0}\nids on screen: \n{1}\n count = {2}".format(frame_num, ids, len(ids))
if d.shape[1]>8 and d.shape[0]>0 and d[0,8] == 1:
# Noisy frame
frame_text = "Noisy Frame\n" + frame_text
draw.text((0, 0), frame_text, (textColor))#(0))
# Custom Offset
dets[:,1] += off_x
dets[:,2] += off_y
# Scale the boxes:
dets[:, 1] -= dets[:, 3] * ((scale_bb - 1)/2)
dets[:, 2] -= dets[:, 4] * ((scale_bb - 1)/2)
dets[:, 3:5] = dets[:, 3:5] * scale_bb
# Convert from [x1, y1, width, height] to [x1, y1, x2, y2]
dets[:, 3:5] += dets[:,1:3]
ids = []
species = {
'-1':"Unknown",
'0':"Human",
'1':"Elephant",
'2':"Lion",
'3':"Giraffe",
'4':"Dog"
}
for i, d in enumerate(dets):
if (show_boxes):
boxcolor = boxColor
if d.shape[0]>7 and d[7] == 1:
boxcolor = 255 - boxColor
draw.rectangle([d[3], d[4], d[1], d[2]], outline=(boxcolor))#c)
#ids.append(d[0])
d = d.astype(np.int32)
if (show_text):
boxTag = str(d[0])
if len(d)>6 and d[5] == 0:
# animal
boxTag += '-' + species[str(d[6])]
draw.text((d[1],d[2] -10),boxTag, (textColor))
if (save_path is not None):
image.save(save_path)
return image
except Exception as e:
#print(str(e))
print("error in redrawing image")
raise e
def get_frame_w_trackers(self, frame_num, scale_bb=1, sort_order='x',
off_x=0, off_y=0, show_boxes=True, boxColor=0,
show_text=True, textColor=0, save_path=None):
img_path = os.path.join(self.img_dir, self.images[frame_num])
det_matrix = {
'GT': self.gt_annotations.get_frame(frame_num)[:,1:].astype(np.float)
}
colors = {
'GT': [textColor, boxColor]
}
index=1
for i in self.tracker_res:
det_matrix[i] = self.tracker_res[i].get_frame(frame_num)[:,1:].astype(np.float)
colors[i] = [COLOR[index], COLOR[index]]
index+=1
try:
with Image.open(img_path) as image:
#print("Mode", image.mode, type(image.mode))
if (image.mode != 'RGB'):
image = image.convert(mode='RGB')
draw = ImageDraw.Draw(image)
frame_text = "Frame: {0}".format(frame_num)
d = det_matrix['GT']
if d.shape[1]>8 and d.shape[0]>0 and d[0,8] == 1:
# Noisy frame
frame_text += " (Noisy Frame)"
draw.text((0, 0), frame_text, (textColor))#(0))
text_start = draw.textsize(frame_text)[1] +1
for e in det_matrix:
dets = det_matrix[e]
textColor = colors[e][0]
boxColor = colors[e][1]
#dets = self.gt_annotations.get_frame(frame_num)[:,1:].astype(np.float)
if len(dets) == 0:
if (save_path is not None):
image.save(save_path)
continue#return image
# Format: [frame, ID, x1, y1, width, height, obj_class, species, occluded, noisy_frame]
labels = None
#print('sort_order', sort_order)
if (sort_order == 'x+y'):
d = dets[np.argsort(dets[:,2] + dets[:,1])]
else:
d = dets[np.lexsort((dets[:,2],dets[:,1]))]
ids = d[:,0]
#print(ids)
h,w = image.size
#frame_text = "Frame: " + frame_num + "ids on screen: \n" + str(ids) + "\n count = " + str(len(ids))
frame_text = "{0}: {1} count = {2}".format(e, ids, len(ids))
draw.text((0, text_start), frame_text, (textColor))#(0))
text_start += draw.textsize(frame_text)[1] +1
# Custom Offset
dets[:,1] += off_x
dets[:,2] += off_y
# Scale the boxes:
dets[:, 1] -= dets[:, 3] * ((scale_bb - 1)/2)
dets[:, 2] -= dets[:, 4] * ((scale_bb - 1)/2)
dets[:, 3:5] = dets[:, 3:5] * scale_bb
# Convert from [x1, y1, width, height] to [x1, y1, x2, y2]
dets[:, 3:5] += dets[:,1:3]
ids = []
species = {
'-1':"Unknown",
'0':"Human",
'1':"Elephant",
'2':"Lion",
'3':"Giraffe",
'4':"Dog"
}
for i, d in enumerate(dets):
if (show_boxes):
boxcolor = boxColor
if d.shape[0]>7 and d[7] == 1:
boxcolor = 255 - boxColor
draw.rectangle([d[3], d[4], d[1], d[2]], outline=(boxcolor))#c)
#ids.append(d[0])
d = d.astype(np.int32)
if (show_text):
boxTag = str(d[0])
if len(d)>6 and d[5] == 0:
# animal
boxTag += '-' + species[str(d[6])]
draw.text((d[1],d[2] -10),boxTag, (textColor))
if (save_path is not None):
image.save(save_path)
return image
except Exception as e:
#print(str(e))
print("error in redrawing image")
raise e
def generate_video(self, output_dir, fps=15, start_frame=0, end_frame=None):
if end_frame is None:
end_frame = len(self.images)
if start_frame > len(self.images) or end_frame < start_frame or end_frame > len(self.images):
print("Invalid input", start_frame, end_frame, "number of images is", len(self.images))
return
outpath = os.path.join(output_dir, self.seqname + '.avi')
out = cv2.VideoWriter(outpath, cv2.VideoWriter_fourcc(*"XVID"), float(fps), self.size)
#for i in tqdm(range(start_frame,end_frame)):
for i in range(start_frame,end_frame):
img_path = './output-imgs/' + self.images[i]
#out.write(np.asarray(self.get_frame(i)))
self.get_frame(i, save_path = img_path)
out.write(cv2.imread(img_path))
out.release()
#print('Written video to', outpath)
def generate_video_w_trackers(self, output_dir, fps=15, start_frame=0, end_frame=None):
if end_frame is None:
end_frame = len(self.images)
if start_frame > len(self.images) or end_frame < start_frame or end_frame > len(self.images):
print("Invalid input", start_frame, end_frame, "number of images is", len(self.images))
return
outpath = os.path.join(output_dir, self.seqname + '.avi')
out = cv2.VideoWriter(outpath, cv2.VideoWriter_fourcc(*"XVID"), float(fps), self.size)
#for i in tqdm(range(start_frame,end_frame)):
for i in range(start_frame,end_frame):
img_path = './output-imgs/' + self.images[i]
#out.write(np.asarray(self.get_frame(i)))
self.get_frame_w_trackers(i, save_path = img_path)
out.write(cv2.imread(img_path))
out.release()
print('Written video to', outpath)
class SotSequence(Sequence):
def __init__(self, orig_image_dir, gt_path, img_list_file, seqname=None, trackers=[]):
if (seqname == None):
# Assumed data format is /path/to/data/seqname.{csv|txt}
self.seqname = gt_path.split('/')[-2]
else:
self.seqname = seqname
super().__init__(orig_image_dir, gt_path, seqname=seqname, trackers=trackers)
with open(img_list_file, 'r') as f:
self.img_map = [i.split(':') for i in f.read().splitlines()]
self.images = [i[1].split('/')[-1] for i in self.img_map]