Edge Detection Using Convolutional Neural Network

In this work, we propose a deep learning method to solve the edge detection problem in image processing area. Existing methods usually rely heavily on computing multiple image features, which makes the whole system complex and computationally expensive. We train Convolutional Neural Networks (CNN) that can make predictions for edges directly from image patches. By adopting such networks, our system is free from additional feature extraction procedures, simple and efficient without losing its detection performance. We also perform experiments on various networks structures, data combination, pre-processing and post-processing techniques, revealing their influence on performance.