# Revisiting Edge Detection in Convolutional Neural Networks

arXiv: https://arxiv.org/abs/2012.13576 · HTML: https://mlsanity.com/paper/2012.13576/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 25 Dec 2020.

## Built on by (most cited first, 8 of 8)

- [DORA: Exploring Outlier Representations in Deep Neural Networks](https://mlsanity.com/paper/2206.04530/) (2022)
- [Boosting Edge Detection with Pixel-wise Feature Selection: The Extractor-Selector Paradigm](https://mlsanity.com/paper/2501.02534/) (2025)
- [A Comprehensive Study for Edge Detection: Ideal-Prior Guidance Learning, High-Precision Architecture, and Better Distinguishability Evaluation Benchmark](https://mlsanity.com/paper/2407.19992/) (2024)
- [Impact of Colour Variation on Robustness of Deep Neural Networks](https://mlsanity.com/paper/2209.02832/) (2022)
- [Efficient Representation of Natural Image Patches](https://mlsanity.com/paper/2210.13004/) (2022)
- [Impact of Light and Shadow on Robustness of Deep Neural Networks](https://mlsanity.com/paper/2305.14165/) (2023)
- [Defective Edge Detection Using Cascaded Ensemble Canny Operator](https://mlsanity.com/paper/2411.14868/) (2024)
- [DFYP: A Dynamic Fusion Framework with Spectral Channel Attention and Adaptive Operator learning for Crop Yield Prediction](https://mlsanity.com/paper/2507.05849/) (2025)

## Abstract (arXiv)

The ability to detect edges is a fundamental attribute necessary to truly capture visual concepts. In this paper, we prove that edges cannot be represented properly in the first convolutional layer of a neural network, and further show that they are poorly captured in popular neural network architectures such as VGG-16 and ResNet. The neural networks are found to rely on color information, which might vary in unexpected ways outside of the datasets used for their evaluation. To improve their robustness, we propose edge-detection units and show that they reduce performance loss and generate qualitatively different representations. By comparing various models, we show that the robustness of edge detection is an important factor contributing to the robustness of models against color noise.