Amped: Adaptive Multi-stage Non-edge Pruning for Edge Detection

2026-03-29

AI Summary

This work addresses the high computational cost and inference latency of Transformer-based edge detection models by proposing an Adaptive Multi-stage Pruning framework for Edge Detection (Amped). Amped introduces, for the first time in edge detection, an adaptive token pruning mechanism that significantly reduces computation by early identification and removal of high-confidence non-edge image tokens. Complementing this strategy, a lightweight and efficient Streamline Edge Detector (SED) is designed to maintain architectural simplicity while achieving strong performance. Evaluated on the BSDS500 dataset, the proposed method achieves up to a 40% reduction in GFLOPs with only a marginal 0.4% drop in F-measure, attaining an ODS score of 86.5%.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Learning to Search

Application Category

• Security and Privacy: Privacy-enhancing technologies
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web

Abstract

Edge detection is a fundamental image analysis task that underpins numerous high-level vision applications. Recent advances in Transformer architectures have significantly improved edge quality by capturing long-range dependencies, but this often comes with computational overhead. Achieving higher pixel-level accuracy requires increased input resolution, further escalating computational cost and limiting practical deployment. Building on the strong representational capacity of recent Transformer-based edge detectors, we propose an Adaptive Multi-stage non-edge Pruning framework for Edge Detection(Amped). Amped identifies high-confidence non-edge tokens and removes them as early as possible to substantially reduce computation, thus retaining high accuracy while cutting GFLOPs and accelerating inference with minimal performance loss. Moreover, to mitigate the structural complexity of existing edge detection networks and facilitate their integration into real-world systems, we introduce a simple yet high-performance Transformer-based model, termed Streamline Edge Detector(SED). Applied to both existing detectors and our SED, the proposed pruning strategy provides a favorable balance between accuracy and efficiency-reducing GFLOPs by up to 40% with only a 0.4% drop in ODS F-measure. In addition, despite its simplicity, SED achieves a state-of-the-art ODS F-measure of 86.5%. The code will be released.

Problem

• edge detection
• computational efficiency
• Transformer
• model pruning
• real-time deployment

Innovation

• non-edge pruning
• Transformer-based edge detection
• computational efficiency
• adaptive multi-stage pruning
• Streamline Edge Detector