Pixel-wise Deep Learning for Contour Detection

TL;DR

This paper presents a pixel-wise deep learning approach for contour detection using DenseNet features and SVM classification, demonstrating effectiveness on the BSDS500 dataset.

Contribution

It introduces a novel combination of multiscale CNN features with SVM for improved contour detection at the pixel level.

Findings

Combining features from different CNN layers enhances detection accuracy.

DenseNet-based features outperform traditional methods.

The approach achieves competitive results on BSDS500.

Abstract

We address the problem of contour detection via per-pixel classifications of edge point. To facilitate the process, the proposed approach leverages with DenseNet, an efficient implementation of multiscale convolutional neural networks (CNNs), to extract an informative feature vector for each pixel and uses an SVM classifier to accomplish contour detection. In the experiment of contour detection, we look into the effectiveness of combining per-pixel features from different CNN layers and verify their performance on BSDS500.