SFGNet: Semantic and Frequency Guided Network for Camouflaged Object Detection

2025-09-14

AI Summary

Existing camouflage object detection (COD) methods overlook semantic disparities among textual prompts and fine-grained frequency-domain features, leading to degraded performance in complex backgrounds and ambiguous boundaries. To address this, we propose a dual-guided COD framework integrating semantic and frequency cues. First, a semantic prompt injection mechanism is introduced to explicitly model discriminative semantics across diverse textual prompts. Second, a multi-band Fourier module (MBFM) is designed to capture multi-scale structural patterns and edge information in the frequency domain. Third, an interactive structure enhancement block (ISEB) is constructed to enable synergistic optimization between spatial- and frequency-domain features. Our method significantly improves structural integrity and boundary-aware accuracy for camouflaged objects. Extensive experiments on three standard COD benchmarks demonstrate consistent superiority over state-of-the-art methods, validating its strong adaptability to complex scenes and robust generalization capability.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Camouflaged object detection (COD) aims to segment objects that blend into their surroundings. However, most existing studies overlook the semantic differences among textual prompts of different targets as well as fine-grained frequency features. In this work, we propose a novel Semantic and Frequency Guided Network (SFGNet), which incorporates semantic prompts and frequency-domain features to capture camouflaged objects and improve boundary perception. We further design Multi-Band Fourier Module(MBFM) to enhance the ability of the network in handling complex backgrounds and blurred boundaries. In addition, we design an Interactive Structure Enhancement Block (ISEB) to ensure structural integrity and boundary details in the predictions. Extensive experiments conducted on three COD benchmark datasets demonstrate that our method significantly outperforms state-of-the-art approaches. The core code of the model is available at the following link: https://github.com/winter794444/SFGNetICASSP2026.

Problem

• Detecting camouflaged objects blending into surroundings
• Incorporating semantic prompts and frequency-domain features
• Handling complex backgrounds and blurred object boundaries

Innovation

• Semantic prompts and frequency features integration
• Multi-Band Fourier Module for complex backgrounds
• Interactive Structure Enhancement Block for boundaries