FCL-COD: Weakly Supervised Camouflaged Object Detection with Frequency-aware and Contrastive Learning

March 2026 Jingchen Ni, Quan Zhang, Dan Jiang, Keyu Lv, Ke Zhang, Chun Yuan CVPR 2026 (CCF-A)
FCL-COD: Weakly Supervised Camouflaged Object Detection with Frequency-aware and Contrastive Learning

Overview

As the first author, I proposed FCL-COD, a frequency-aware and contrastive learning-based weakly-supervised COD framework. It incorporates Frequency-aware Low-rank Adaptation (FoRA) into SAM to suppress non-camouflage responses, and employs gradient-aware contrastive learning with multi-scale frequency-aware representation learning to achieve precise boundary delineation. The method surpasses both state-of-the-art weakly-supervised and fully-supervised techniques.