CATP: Confidence-Aware Token Pruning for Camouflaged Object Detection

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Gao, Yuhan, Kang, Shuhao, He, Xin, Li, Bing, Cheng, Xu, Liu, Yun
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911604003569664
author Gao, Yuhan
Kang, Shuhao
He, Xin
Li, Bing
Cheng, Xu
Liu, Yun
author_facet Gao, Yuhan
Kang, Shuhao
He, Xin
Li, Bing
Cheng, Xu
Liu, Yun
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2604_16854
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CATP: Confidence-Aware Token Pruning for Camouflaged Object Detection
Gao, Yuhan
Kang, Shuhao
He, Xin
Li, Bing
Cheng, Xu
Liu, Yun
Computer Vision and Pattern Recognition
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.
title CATP: Confidence-Aware Token Pruning for Camouflaged Object Detection
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.16854