-
Notifications
You must be signed in to change notification settings - Fork 0
/
rknn_tf_forward.py
157 lines (121 loc) · 5.29 KB
/
rknn_tf_forward.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
import numpy as np
import torch
import torch.nn as nn
import cv2
import tensorflow as tf
from packaging import version
from model import PoseHighResolutionNet, extra32
if version.parse(tf.__version__) < version.parse("2.6"):
import tensorflow.keras as keras
from tensorflow.keras import layers
else:
import keras
from keras import layers
from tensorflow_model_optimization.python.core.quantization.keras.quantize_wrapper import QuantizeWrapper
from model_compression_toolkit.qat.keras.quantizer.configs.weight_quantizer_config import WeightQuantizeConfig
from dataset import prepare
import tensorflow_datasets as tfds
from loss import AELoss
def compare(a,b):
print((a-b).min(), (a-b).max())
print((a-b).sum()/128/128/2)
if __name__ == "__main__":
index2label = {0: "label1", 1:"label2"}
# load image
# img = cv2.imread("/home/fsw/Documents/codes/CenterNet_match/snapshot/foot_robot1028/src_img.jpg")
img = cv2.imread("/home/fsw/Documents/codes/CenterNet_match/dataset/foot_robot/train/10.jpg")
img_h, img_w, _ = img.shape
input_w, input_h = 512, 512
image = cv2.resize(img, (512, 512),cv2.INTER_NEAREST)
mean = np.array([0.408, 0.447, 0.47], dtype=np.float32) * 255
std = np.array([0.289, 0.274, 0.278], dtype=np.float32) * 255
image = (image - mean) / std
image = np.expand_dims(image, axis=0)
model_path = r'foot_robot1028_512_512_160.pth'
model = get_hr(num_layers=32, heads={"hm": 2, "em": 2, "reg": 2}).cuda()
checkpoint = torch.load(model_path, map_location="cpu")
new_checkpoint = collections.OrderedDict()
for name, module in checkpoint.items():
name = name.replace("module.", "")
new_checkpoint[name] = module
model.load_state_dict(new_checkpoint, strict=True)
model.cuda()
model.eval()
def nms(heat, kernel=1):
pad = (kernel - 1) // 2
hmax = nn.functional.max_pool2d(heat, (kernel, kernel), stride=1, padding=pad)
keep = (hmax == heat).float()
return heat * keep, hmax
def topk(scores, K=20):
batch, cat, height, width = scores.size()
topk_scores, topk_inds = torch.topk(scores.contiguous().view(batch, -1), K)
topk_clses = torch.floor_divide(topk_inds, (height * width)).int()
topk_inds = topk_inds % (height * width)
topk_ys = torch.floor_divide(topk_inds, width).int().float()
topk_xs = (topk_inds % width).int().float()
return topk_scores, topk_inds, topk_clses, topk_ys, topk_xs
nmskey, hmax = nms(hm, 3)
kscore, kinds, kcls, kys, kxs = topk(nmskey, K=10)
kys = kys.cpu().data.numpy().astype(np.int)
kxs = kxs.cpu().data.numpy().astype(np.int)
key = [[], []] # key是目标点在ht中的坐标
good_scores = []
good_cls = []
for ind in range(kscore.shape[1]):
score = kscore[0, ind]
cls = kcls[0, ind]
# if score > 0.05:
if score > 0.3:
key[0].append(kys[0, ind])
key[1].append(kxs[0, ind])
good_scores.append(score)
good_cls.append(cls)
stride = 4
if key[0] is not None and len(key[0]) > 0:
pred_centers = np.zeros([2, len(good_cls)])
pre_centers_em = np.zeros([len(good_cls)])
for j in range(len(good_cls)):
# 类别
cls = good_cls[j]
# 中心点
print(key[0][j], key[1][j])
reg_x = xy[0, 0, key[0][j], key[1][j]]
reg_y = xy[0, 1, key[0][j], key[1][j]]
cx = (reg_x + key[1][j]) * stride
cy = (reg_y + key[0][j]) * stride
pred_centers[0, j] = cx
pred_centers[1, j] = cy
# embedding
pre_centers_em[j] = em[0, cls.item(), key[0][j], key[1][j]]
# print(pre_centers_em[j])
# oem1 = reg_em1 * stride
# oem2 = reg_em2 * stride
else:
pred_centers = (None, None, None, None)
pre_centers_em = None
x, y = pred_centers
if x is not None:
# image = restoreImage(images, 0)
# image = image.astype(np.uint8).copy()
scale_w, scale_h = input_w / img_w, input_h / img_h
for i in range(x.shape[0]):
x_ = max((1 / scale_w) * x[i], 0)
y_ = max((1 / scale_h) * y[i], 0)
if good_scores[i] > 0.3:
cv2.circle(img, (int(x_), int(y_)), 4, (255, 0, 255), -1)
cv2.putText(img, index2label[int(good_cls[i])], (int(x_ + 10), int(y_ + 20)),
cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 255), 1)
# cv2.putText(img, str(round(float(good_scores[i]), 3)), (int(x_ + 30), int(y_ + 20)),
# cv2.FONT_HERSHEY_SIMPLEX, 0.8, (20, 240, 0), 1)
cv2.putText(img, str(round(float(pre_centers_em[i]), 3)), (int(x_ + 30), int(y_ + 20)),
cv2.FONT_HERSHEY_SIMPLEX, 0.8, (20, 240, 0), 1)
# txt_f.write(' '.join(
# [index2label[int(good_cls[i])], str(int(x_)), str(int(y_)), str(float(good_scores[i]))]) + '\n')
# print(os.path.join(img_save_dir, img_name))
cv2.imshow("img",img)
cv2.waitKey()
# cv2.imwrite('./onnx_test_lq1117_2.jpg', image_source)
# # perf
# print('--> Begin evaluate model performance')
# perf_results = rknn.eval_perf(inputs=[img])
# print('done')