A Brief Analysis of the Holistically-Nested Edge Detector

This work describes the HED method for edge detection. uses a neural network based on VGG16 backbone, supplemented with some extra layers merging results at different scales. The training was performed an augmented version of BSDS500 dataset. We perform brief analysis produced by HED, highlighting its quality but also indicating limitations. Overall, produces state-of-the-art results. **This is MLBriefs article, source code has not been reviewed!** **The original [[available here|https://github.com/sniklaus/pytorch-hed/tree/8db09037f2491abbed4e5d7e37ec75210a2dbfbf]] (last checked 2022/10/03).**