RHN: A Residual Holistic Neural Network for Edge Detection

Edge detection plays a very important role in many image processing and computer vision applications. Use of deep convolutional neural networks (DCNNs) has significantly advanced the performance of image edge detection techniques. Existing DCNN techniques, which make use of residual learning, exhibit a good edge detection performance at the expense of an extremely high computational complexity. There are a few VGG16-based DCNN techniques for edge detection that have been proposed with relatively much lower complexity. In this paper, by using the mechanism of residual learning, a new VGG16-based DCNN technique for edge detection is proposed with a view to provide a performance superior to that provided by other such networks while still preserving their low complexity. The proposed network is experimented on different datasets and is shown to outperform all the other VGG16-based techniques designed to solve the problem of edge detection.

Automated summary

In this paper, a new VGG16-based DCNN technique for edge detection is proposed, utilizing residual learning mechanism to achieve superior performance while maintaining low complexity. Experimental results demonstrate that the proposed network outperforms other VGG16-based techniques on different datasets.