-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathtest.py
168 lines (142 loc) · 5.86 KB
/
test.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
# -*- coding: utf-8 -*-
# @Time : 2020/3/30 13:16
# @Author : Suke0
# @Email : [email protected]
# @File : test.py
# @Software: PyCharm
import cv2
import numpy as np
from PIL import Image, ImageDraw
from anchor import two_boxes_iou, anchors_for_shape, bbox_transform_env
from image import random_visual_effect_generator, resize_image, read_image_bgr, preprocess_image
from model import create_model
# LABELS = ["aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat",
# "chair", "cow", "diningtable", "dog", "horse", "motorbike", "person",
# "pottedplant", "sheep", "sofa", "train", "tvmonitor"
# ]
from weight_npy_transfor_h5 import transfor_weight
LABELS = ['person','bicycle','car','motorbike','aeroplane','bus','train','truck','boat','traffic light','fire hydrant','stop sign','parking meter',
'bench','bird','cat','dog','horse','sheep','cow','elephant','bear','zebra','giraffe','backpack','umbrella','handbag','tie','suitcase',
'frisbee','skis','snowboard','sports ball','kite','baseball bat','baseball glove','skateboard','surfboard','tennis racket','bottle',
'wine glass','cup', 'fork','knife','spoon','bowl','banana','apple','sandwich','orange','broccoli','carrot','hot dog','pizza','donut',
'cake','chair','sofa','pottedplant','bed','diningtable','toilet','tvmonitor','laptop','mouse','remote','keyboard','cell phone',
'microwave','oven','toaster','sink','refrigerator','book','clock','vase','scissors','teddy bear','hair drier','toothbrush'
]
def preprocess_img(img_path):
image = np.asarray(Image.open(img_path).convert('RGB'))
image = image[:, :, ::-1].copy()
# visual_effect_generator = random_visual_effect_generator(
# contrast_range=(0.9, 1.1),
# brightness_range=(-.1, .1),
# hue_range=(-0.05, 0.05),
# saturation_range=(0.95, 1.05)
# )
#
# visual_effect = next(visual_effect_generator)
# # apply visual effect
# image = visual_effect(image_bgr)
image = image.astype(np.float32)
image -= [103.939, 116.779, 123.68]
# resize image
image, image_scale = resize_image(image)
return image, image_scale
pass
#使用NMS方法，对结果去重
def non_max_suppression(bbox_pred,cls_pred, anchors, confidence_threshold=0.5, iou_threshold=0.4):
# 过滤掉概率小于0.5的预测值
idxs = np.argmax(cls_pred, axis=-1)
indexs = range(cls_pred.shape[0])
cls_pred = cls_pred[indexs, idxs]
t_index = np.where(cls_pred > confidence_threshold)
anchors = anchors[t_index]
cls_pred = np.expand_dims(cls_pred, -1)
cls_pred = cls_pred[t_index]
labels_ = idxs[t_index]
labels_ = np.unique(labels_)
idxs = np.expand_dims(idxs, -1)
labels = idxs[t_index]
bbox_pred = bbox_pred[t_index]
bbox_pred = bbox_transform_env(anchors,bbox_pred)
predictions = np.concatenate([labels, cls_pred, bbox_pred], axis=-1)
result = []
# print(f'正例样本数：{len(predictions)}')
for label in labels_:
idxs = predictions[:, 0] == label
label_pred_boxes = predictions[idxs]
while len(label_pred_boxes) > 0:
idxs = np.argsort(-label_pred_boxes[:, 1]) # 降序排序
label_max_box = label_pred_boxes[idxs[0]]
label_pred_boxes = label_pred_boxes[idxs[1:]]
result.append(label_max_box)
box1 = label_max_box[2:6]
for i, box2 in enumerate(label_pred_boxes[:, 2:6]):
iou = two_boxes_iou(box1, box2)
if iou > iou_threshold:
label_pred_boxes[i, 0] = -1
pass
label_pred_boxes = label_pred_boxes[label_pred_boxes[:, 0] > -1]
if len(label_pred_boxes) == 1:
label_pred_boxes = np.reshape(label_pred_boxes, (1, 6))
pass
pass
return np.array(result) # (n_boxes, 1+1+4)
pass
# 将级别结果显示在图片上
def draw_boxes(boxes, img_file, cls_names, img_scale):
img = Image.open(img_file)
draw = ImageDraw.Draw(img)
for box in boxes:
box[2:] = box[2:] / img_scale
draw.rectangle(list(box[2:]), outline='red')
draw.text(list(box[2:4]),'{} {:.2f}%'.format(cls_names[int(box[0])], box[1] * 100), fill='red')
print('{} {:.2f}%'.format(cls_names[int(box[0])], box[1] * 100), list(box[2:]))
img.save(f"output_img.jpg")
img.show()
pass
def predict(img_path,cls_names = LABELS):
# # load image
# image = read_image_bgr(img_path)
#
# # copy to draw on
# draw = image.copy()
# draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
#
# # preprocess image for network
# image = preprocess_image(image)
# image, img_scale = resize_image(image)
model = create_model(80)
model.load_weights("./weight/resnet50retinanet_coco.h5")
#transfor_weight(model)
vars = model.variables
# weights = np.load('./weight/weight.npy', allow_pickle=True)
# arr = []
print("---------------------------------------------")
for v in vars:
print(v.name + '__' + str(v.shape))
pass
#
# for v in vars:
# vname = v.name.split('/')[-1]
# if 'kernel' not in vname:
# print(vname + '__' + str(v.shape))
# arr.append(v)
# pass
# pass
# print("-------------------------")
# print(arr[-1])
# print("-------------------------")
# print(weights[-1])
# print("-------------------------")
image, img_scale = preprocess_img(img_path)
anchors = anchors_for_shape(image.shape)
#测试单张图片
img = np.expand_dims(image,0)
bbox_pred, cls_pred = model(img)
bbox_pred = np.squeeze(bbox_pred)
cls_pred = np.squeeze(cls_pred)
bboxes = non_max_suppression(bbox_pred, cls_pred, anchors)
draw_boxes(bboxes,img_path,cls_names,img_scale)
pass
if __name__ == '__main__':
predict('./test_img/2004374.jpg')
pass