-
Notifications
You must be signed in to change notification settings - Fork 0
/
feature_matching_for_traing.py
192 lines (141 loc) · 7.18 KB
/
feature_matching_for_traing.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
import os
import cv2
import numpy as np
import time
import argparse
from pathlib import Path
import torch
from torch import nn
from lightglue import LightGlue, SuperPoint, DISK, SIFT, ALIKED, DoGHardNet
from lightglue.utils import load_image, rbd
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
def makedir(d):
if not os.path.exists(d):
os.makedirs(d)
class Feature():
def __init__(self,
feature_num = 512,
):
self.feature_num = feature_num
self.device = 'cuda' if torch.cuda.is_available() else 'cpu'
self.detector = SuperPoint(max_num_keypoints=feature_num).eval().cuda()
self.matcher = LightGlue(features='superpoint').eval().cuda() # load the matcher
def detect(self, image_path):
image = load_image(image_path).cuda()
features = self.detector.extract(image)
return features
def match(self, feats0, feats1):
matches01 = self.matcher({'image0': feats0, 'image1': feats1})
matches01 = rbd(matches01)
return matches01['matches'].cpu()
def find_inliers(kpts1, kpts2):
if len(kpts1) < 10 or len(kpts2) < 10:
return None
F, mask = cv2.findFundamentalMat(kpts1, kpts2, cv2.FM_RANSAC, 20)
mask = mask.reshape(-1) if (mask is not None and np.sum(mask) > 100) else None
return mask
def draw_keypoints(image, kpts):
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
result_img = cv2.cvtColor(gray, cv2.COLOR_GRAY2RGB)
for pt1 in kpts:
cv2.circle(result_img, tuple(pt1), 2, (0, 255, 0), -1, lineType=cv2.LINE_AA) # 绿色，实心，半径为3
return result_img
def draw_matches(last_image, image, last_kpts, kpts, track_ids, save_path):
drawed_img1 = draw_keypoints(last_image, last_kpts)
drawed_img2 = draw_keypoints(image, kpts)
h1, w1, _ = drawed_img1.shape
h2, w2, _ = drawed_img2.shape
result_img = np.ones((max(h1, h2), w1 + w2 + 10, 3), dtype=np.uint8) * 255
result_img[:h1, :w1] = drawed_img1
result_img[:h2, w1 + 10:] = drawed_img2
for i, tid1 in enumerate(track_ids):
if tid1 == -1:
continue # 跳过没有track的点
pt1 = tuple(last_kpts[tid1])
pt1 = tuple(round(x) for x in pt1)
pt2 = kpts[i]
pt2 = tuple([pt2[0] + w1 + 10, pt2[1]])
pt2 = tuple(round(x) for x in pt2)
cv2.line(result_img, pt1, pt2, (0, 255, 0), 1, lineType=cv2.LINE_AA) # 绿色，粗2个像素，50%透明
overlay = result_img.copy()
cv2.addWeighted(overlay, 0.5, result_img, 1 - 0.5, 0, result_img) # 设置透明度为0.5
cv2.imwrite(save_path, result_img)
def process_sequence(seq_root, feature):
rgb_root = os.path.join(seq_root, "rgb")
tracking_root = os.path.join(seq_root, "matching")
pose_root = os.path.join(seq_root, "poses")
cali_root = os.path.join(seq_root, "calibration")
matching_vis_root = os.path.join(seq_root, "matching_vis")
makedir(tracking_root)
makedir(matching_vis_root)
image_names = os.listdir(rgb_root)
image_names.sort()
ref_id, num_since_last_ref = None, 0
last_keypoints, last_track_ids, last_pose, last_image, last_K = None, None, None, None, None
for i in range(len(image_names)):
image_name = image_names[i]
image_path = os.path.join(rgb_root, image_name)
image = cv2.imread(image_path)
image_idx = image_name.split('.')[0]
tracking_file = os.path.join(tracking_root, image_idx + ".matching.txt")
add_new_keyframe = (i == 0)
keypoints = feature.detect(image_path)
kpts = rbd(keypoints)['keypoints'].cpu().numpy()
M = len(kpts)
print("M = {}, image_path = {}".format(M, image_path))
if not add_new_keyframe:
matches = feature.match(last_keypoints, keypoints)
last_kpts = rbd(last_keypoints)['keypoints'].cpu().numpy()
inliers = find_inliers(last_kpts[matches[:, 0]], kpts[matches[:, 1]])
if inliers is not None:
matches = matches[inliers]
parallax = last_kpts[matches[:, 0]] - kpts[matches[:, 1]]
avg_parallax = np.mean(np.linalg.norm(parallax, axis=1))
num_tracked = len(matches)
track_ids = np.full(M, -1)
track_ids[matches[:, 1]] = matches[:, 0]
# save_path = os.path.join(matching_vis_root, image_name)
# draw_matches(last_image, image, last_kpts, kpts, track_ids, save_path)
add_new_keyframe = (num_tracked < 0.5 * len(last_kpts)) or (avg_parallax > 0.15 * min(image.shape[0], image.shape[1]))
else:
add_new_keyframe = True
if add_new_keyframe:
track_ids = [id for id in range(M)]
track_ids = np.array(track_ids)
ref_id = i
last_keypoints = keypoints
last_track_ids = track_ids
last_image = image
num_since_last_ref = 0
num_since_last_ref += 1
meta_data = np.array([i, ref_id, M])
ids_and_kpts = np.hstack((track_ids[:, np.newaxis], kpts))
save_info = np.vstack((meta_data[np.newaxis, :], ids_and_kpts))
np.savetxt(tracking_file, save_info)
if __name__ == '__main__':
parser = argparse.ArgumentParser(
description='Image feature matching for a specific sequence is processed through a specified list of data paths and scenes.',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('dataroot', type=Path,
help='The root catalog of the dataset, e.g. "/home/user/project/datasets"')
parser.add_argument('sequences', type=str, nargs='+',
help='A list of sequences to be processed, e.g. "pgt_7scenes_chess pgt_7scenes_heads"')
args = parser.parse_args()
# dataroot = "/home/xukuan/project/seq_ace/seq_scr/datasets"
# sequences = ['pgt_7scenes_chess', 'pgt_7scenes_heads', 'pgt_7scenes_pumpkin', 'pgt_7scenes_fire', 'pgt_7scenes_office', 'pgt_7scenes_redkitchen', 'pgt_7scenes_stairs']
# sequences = ['Cambridge_GreatCourt', 'Cambridge_KingsCollege', 'Cambridge_OldHospital', 'Cambridge_ShopFacade', 'Cambridge_StMarysChurch']
# sequences = ['wayspots_bears', 'wayspots_cubes', 'wayspots_inscription', 'wayspots_lawn', 'wayspots_map', 'wayspots_squarebench', 'wayspots_statue', 'wayspots_tendrils', 'wayspots_therock', 'wayspots_wintersign']
# sequences = ['pgt_12scenes_apt1_kitchen', 'pgt_12scenes_apt1_living', 'pgt_12scenes_apt2_bed',
# 'pgt_12scenes_apt2_kitchen', 'pgt_12scenes_apt2_living', 'pgt_12scenes_apt2_luke',
# 'pgt_12scenes_office1_gates362', 'pgt_12scenes_office1_gates381', 'pgt_12scenes_office1_lounge',
# 'pgt_12scenes_office1_manolis', 'pgt_12scenes_office2_5a', 'pgt_12scenes_office2_5b']
time0 = time.time()
train_feature_num = 1000
train_feature = Feature(train_feature_num)
for seq in args.sequences:
print("processing {} ....".format(seq))
seq_root = args.dataroot / seq
traing_root = seq_root / "train"
process_sequence(str(traing_root), train_feature)
time1 = time.time()
print("time = {}".format(time1 - time0))