-
Notifications
You must be signed in to change notification settings - Fork 1
/
detect.py
55 lines (44 loc) · 2.33 KB
/
detect.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
# coding=utf-8
import cv2
import argparse
from models import *
from PIL import Image
import os
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--data', type=str, default='cfg/mask.data', help='*.data file path')
parser.add_argument('--img_size', type=int, default=608, help='inference size (pixels)')
parser.add_argument('--weights', type=str, default='./weights/darknet/yolov3', help='initial weights')
parser.add_argument('--input_path', type=str, default='./data/input/', help='input_path')
parser.add_argument('--iou_loss_thresh', type=int, default=0.5, help='iou_loss_thresh')
parser.add_argument('--confidence_thresh', type=int, default=0.5, help='confidence_thresh')
parser.add_argument('--nms_thresh', type=int, default=0.45, help='nms_thresh')
parser.add_argument('--model', type=str, default='darknet', help='initial weights')
opt = parser.parse_args()
print(opt)
visible_gpus = tf.config.experimental.list_physical_devices('GPU')
print('Visible devices : ', visible_gpus)
model = yolov3(opt)
input_layer = tf.keras.layers.Input([opt.img_size, opt.img_size, 3])
output = model(input_layer, training=False)
if str(opt.weights).split('/')[2] == 'init':
utils.load_weights(model, opt.weights)
else:
model.load_weights(opt.weights)
model.summary()
input_path = os.listdir(opt.input_path)
for image_path in input_path:
image_path = opt.input_path + image_path
original_image = cv2.imread(image_path)
original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
original_image_size = original_image.shape[:2]
image_data = utils.image_preporcess(np.copy(original_image), [opt.img_size, opt.img_size])
image_data = image_data[np.newaxis, ...].astype(np.float32)
pred_bbox = model(image_data, training=False)
pred_bbox = [tf.reshape(x, (-1, tf.shape(x)[-1])) for x in pred_bbox]
pred_bbox = tf.concat(pred_bbox, axis=0)
bboxes = utils.postprocess_boxes(pred_bbox, original_image_size, opt.img_size, opt.confidence_thresh)
bboxes = utils.nms(bboxes, opt.nms_thresh, method='nms')
image = utils.draw_bbox(original_image, bboxes, opt=opt, )
image = Image.fromarray(image)
image.save(image_path.replace('input', 'output'))