JeVoisBase  1.21
JeVois Smart Embedded Machine Vision Toolkit Base Modules
Share this page:
Loading...
Searching...
No Matches
demo.py
Go to the documentation of this file.
1# This file is part of OpenCV Zoo project.
2# It is subject to the license terms in the LICENSE file found in the same directory.
3#
4# Copyright (C) 2021, Shenzhen Institute of Artificial Intelligence and Robotics for Society, all rights reserved.
5# Third party copyrights are property of their respective owners.
6
7import argparse
8
9import numpy as np
10import cv2 as cv
11
12from db import DB
13
14def str2bool(v):
15 if v.lower() in ['on', 'yes', 'true', 'y', 't']:
16 return True
17 elif v.lower() in ['off', 'no', 'false', 'n', 'f']:
18 return False
19 else:
20 raise NotImplementedError
21
22backends = [cv.dnn.DNN_BACKEND_OPENCV, cv.dnn.DNN_BACKEND_CUDA]
23targets = [cv.dnn.DNN_TARGET_CPU, cv.dnn.DNN_TARGET_CUDA, cv.dnn.DNN_TARGET_CUDA_FP16]
24help_msg_backends = "Choose one of the computation backends: {:d}: OpenCV implementation (default); {:d}: CUDA"
25help_msg_targets = "Chose one of the target computation devices: {:d}: CPU (default); {:d}: CUDA; {:d}: CUDA fp16"
26try:
27 backends += [cv.dnn.DNN_BACKEND_TIMVX]
28 targets += [cv.dnn.DNN_TARGET_NPU]
29 help_msg_backends += "; {:d}: TIMVX"
30 help_msg_targets += "; {:d}: NPU"
31except:
32 print('This version of OpenCV does not support TIM-VX and NPU. Visit https://gist.github.com/fengyuentau/5a7a5ba36328f2b763aea026c43fa45f for more information.')
33
34parser = argparse.ArgumentParser(description='Real-time Scene Text Detection with Differentiable Binarization (https://arxiv.org/abs/1911.08947).')
35parser.add_argument('--input', '-i', type=str, help='Path to the input image. Omit for using default camera.')
36parser.add_argument('--model', '-m', type=str, default='text_detection_DB_TD500_resnet18_2021sep.onnx', help='Path to the model.')
37parser.add_argument('--backend', '-b', type=int, default=backends[0], help=help_msg_backends.format(*backends))
38parser.add_argument('--target', '-t', type=int, default=targets[0], help=help_msg_targets.format(*targets))
39parser.add_argument('--width', type=int, default=736,
40 help='Preprocess input image by resizing to a specific width. It should be multiple by 32.')
41parser.add_argument('--height', type=int, default=736,
42 help='Preprocess input image by resizing to a specific height. It should be multiple by 32.')
43parser.add_argument('--binary_threshold', type=float, default=0.3, help='Threshold of the binary map.')
44parser.add_argument('--polygon_threshold', type=float, default=0.5, help='Threshold of polygons.')
45parser.add_argument('--max_candidates', type=int, default=200, help='Max candidates of polygons.')
46parser.add_argument('--unclip_ratio', type=np.float64, default=2.0, help=' The unclip ratio of the detected text region, which determines the output size.')
47parser.add_argument('--save', '-s', type=str, default=False, help='Set true to save results. This flag is invalid when using camera.')
48parser.add_argument('--vis', '-v', type=str2bool, default=True, help='Set true to open a window for result visualization. This flag is invalid when using camera.')
49args = parser.parse_args()
50
51def visualize(image, results, box_color=(0, 255, 0), text_color=(0, 0, 255), isClosed=True, thickness=2, fps=None):
52 output = image.copy()
53
54 if fps is not None:
55 cv.putText(output, 'FPS: {:.2f}'.format(fps), (0, 15), cv.FONT_HERSHEY_SIMPLEX, 0.5, text_color)
56
57 pts = np.array(results[0])
58 output = cv.polylines(output, pts, isClosed, box_color, thickness)
59
60 return output
61
62if __name__ == '__main__':
63 # Instantiate DB
64 model = DB(modelPath=args.model,
65 inputSize=[args.width, args.height],
66 binaryThreshold=args.binary_threshold,
67 polygonThreshold=args.polygon_threshold,
68 maxCandidates=args.max_candidates,
69 unclipRatio=args.unclip_ratio,
70 backendId=args.backend,
71 targetId=args.target
72 )
73
74 # If input is an image
75 if args.input is not None:
76 image = cv.imread(args.input)
77 image = cv.resize(image, [args.width, args.height])
78
79 # Inference
80 results = model.infer(image)
81
82 # Print results
83 print('{} texts detected.'.format(len(results[0])))
84 for idx, (bbox, score) in enumerate(zip(results[0], results[1])):
85 print('{}: {} {} {} {}, {:.2f}'.format(idx, bbox[0], bbox[1], bbox[2], bbox[3], score))
86
87 # Draw results on the input image
88 image = visualize(image, results)
89
90 # Save results if save is true
91 if args.save:
92 print('Resutls saved to result.jpg\n')
93 cv.imwrite('result.jpg', image)
94
95 # Visualize results in a new window
96 if args.vis:
97 cv.namedWindow(args.input, cv.WINDOW_AUTOSIZE)
98 cv.imshow(args.input, image)
99 cv.waitKey(0)
100 else: # Omit input to call default camera
101 deviceId = 0
102 cap = cv.VideoCapture(deviceId)
103
104 tm = cv.TickMeter()
105 while cv.waitKey(1) < 0:
106 hasFrame, frame = cap.read()
107 if not hasFrame:
108 print('No frames grabbed!')
109 break
110
111 frame = cv.resize(frame, [args.width, args.height])
112 # Inference
113 tm.start()
114 results = model.infer(frame) # results is a tuple
115 tm.stop()
116
117 # Draw results on the input image
118 frame = visualize(frame, results, fps=tm.getFPS())
119
120 # Visualize results in a new Window
121 cv.imshow('{} Demo'.format(model.name), frame)
122
123 tm.reset()
124
Definition db.py:10
visualize(image, results, box_color=(0, 255, 0), text_color=(0, 0, 255), fps=None)
Definition demo.py:46