CATP: Confidence-Aware Token Pruning for Camouflaged Object Detection

2026-04-18

AI Summary

This work addresses the high computational cost and deployment challenges of Transformer-based camouflaged object detection (COD) models by proposing a hierarchical confidence-aware token pruning framework. The method dynamically prunes redundant tokens in background and interior regions by evaluating their confidence scores, thereby concentrating computational resources on critical boundary areas. To mitigate information loss induced by pruning, a dual-path feature compensation mechanism is introduced. This study presents the first integration of confidence-aware hierarchical pruning with dual-path compensation, achieving significant reductions in computational complexity across multiple COD benchmarks while maintaining high detection accuracy—thus effectively balancing efficiency and performance.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT

Abstract

Camouflaged Object Detection (COD) aims to segment targets that share extreme textural and structural similarities with their complex environments. Leveraging their capacity for long-range dependency modeling, Transformer-based detectors have become the mainstream approach and achieve state-of-the-art (SoTA) accuracy, yet their substantial computational overhead severely limits practical deployment. To address this, we propose a hierarchical Confidence-Aware Token Pruning framework (CATP) tailored for COD. Our approach hierarchically identifies and discards easily distinguishable tokens from both background and object interiors, focusing computations on critical boundary tokens. To compensate for information loss from pruning, we introduce a dual-path feature compensation mechanism that aggregates contextual knowledge from pruned tokens into enriched features. Extensive experiments on multiple COD benchmarks demonstrate that our method significantly reduces computational complexity while maintaining high accuracy, offering a promising research direction for the efficient deployment of COD models in real-world scenarios. The code will be released.

Problem

• Camouflaged Object Detection
• Transformer
• Computational Overhead
• Model Deployment
• Efficiency

Innovation

• Token Pruning
• Camouflaged Object Detection
• Confidence-Aware
• Transformer Efficiency
• Feature Compensation