Multi-Scale Tensorial Summation and Dimensional Reduction Guided Neural Network for Edge Detection

2025-04-22

AI Summary

To address inaccurate edge localization caused by excessive redundancy and limited receptive fields in edge detection, this paper proposes MTS-DR-Net. It introduces the MTS-DR module—the first integration of a multi-scale tensor summation (MTS) operator with a learnable dimensionality reduction mechanism—enabling large-receptive-field modeling and critical subspace focusing at shallow network layers. Coupled with a U-shaped weighted refinement architecture, the framework enhances edge localization precision. As an end-to-end trainable model, MTS-DR-Net overcomes the conventional CNN limitation of relying on deep stacking to expand receptive fields. Evaluated on BSDS500 and BIPEDv2, it achieves state-of-the-art performance, improving F-measure by 1.8% and 2.3%, respectively, while reducing parameter count by 37% and accelerating inference by 2.1×.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web

Abstract

Edge detection has attracted considerable attention thanks to its exceptional ability to enhance performance in downstream computer vision tasks. In recent years, various deep learning methods have been explored for edge detection tasks resulting in a significant performance improvement compared to conventional computer vision algorithms. In neural networks, edge detection tasks require considerably large receptive fields to provide satisfactory performance. In a typical convolutional operation, such a large receptive field can be achieved by utilizing a significant number of consecutive layers, which yields deep network structures. Recently, a Multi-scale Tensorial Summation (MTS) factorization operator was presented, which can achieve very large receptive fields even from the initial layers. In this paper, we propose a novel MTS Dimensional Reduction (MTS-DR) module guided neural network, MTS-DR-Net, for the edge detection task. The MTS-DR-Net uses MTS layers, and corresponding MTS-DR blocks as a new backbone to remove redundant information initially. Such a dimensional reduction module enables the neural network to focus specifically on relevant information (i.e., necessary subspaces). Finally, a weight U-shaped refinement module follows MTS-DR blocks in the MTS-DR-Net. We conducted extensive experiments on two benchmark edge detection datasets: BSDS500 and BIPEDv2 to verify the effectiveness of our model. The implementation of the proposed MTS-DR-Net can be found at https://github.com/LeiXuAI/MTS-DR-Net.git.

Problem

• Develops MTS-DR-Net for edge detection with large receptive fields
• Reduces redundant information using MTS-DR blocks for efficiency
• Enhances edge detection accuracy on BSDS500 and BIPEDv2 datasets

Innovation

• Multi-scale Tensorial Summation for large receptive fields
• MTS Dimensional Reduction to remove redundancy
• U-shaped refinement module for edge detection