神经网络边缘检测hed

 

 

整体嵌套的边缘检测:

Holistically-Nested Edge Detection

 

vgg16:

https://github.com/harsimrat-eyeem/holy-edge

 

c++版的:

https://github.com/harsimrat-eyeem/hed

 

 

需要建一个环境变量:OMP_NUM_THREADS

os.environ.get('OMP_NUM_THREADS')

也可以不用建环境变量,可以直接改为4

 

建默认参数:

if __name__ == '__main__':

# args.run_test=1
# args.config_file="hed/configs/hed.yaml"
# args.gpu_limit=0.4
# args.download_data=1
parser = argparse.ArgumentParser(description='Utility for Training/Testing DL models(Concepts/Captions) using theano/keras')
parser.add_argument('--config-file', dest='config_file', type=str,default="hed/configs/hed.yaml", help='Experiment configuration file')
parser.add_argument('--train', dest='run_train', action='store_true', default=1, help='Launch training')
parser.add_argument('--test', dest='run_test', action='store_true', default= 0 , help='Launch testing on a list of images')
parser.add_argument('--download-data', dest='download_data', action='store_true', default=False, help='Download training data')
parser.add_argument('--gpu-limit', dest='gpu_limit', type=float, default=1.0, help='Use fraction of GPU memory (Useful with TensorFlow backend)')

 

你可能感兴趣的:(深度学习,视觉相关)