Uncertainty-Masked Bernoulli Diffusion for Camouflaged Object Detection Refinement

2025-06-12

AI Summary

To address low segmentation accuracy in camouflaged object detection (COD) caused by high visual similarity between objects and backgrounds, this paper proposes the first generative post-processing refinement framework tailored for COD. Methodologically, it introduces a hybrid uncertainty quantification network (HUQNet) that jointly models epistemic and aleatoric uncertainties via a Bernoulli diffusion mechanism, enabling adaptive, selective refinement of low-confidence residual regions. Uncertainty masks guide the diffusion sampling process, while a lightweight encoder-decoder–compatible architecture ensures plug-and-play deployment. Evaluated on multiple COD benchmarks, the framework achieves an average 5.5% reduction in mean absolute error (MAE) and a 3.2% improvement in weighted F-measure, with negligible computational overhead. This work establishes a novel uncertainty-aware generative paradigm for COD refinement, significantly boosting the performance of existing COD models without architectural modifications.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Diffusion Models for Vision
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Camouflaged Object Detection (COD) presents inherent challenges due to the subtle visual differences between targets and their backgrounds. While existing methods have made notable progress, there remains significant potential for post-processing refinement that has yet to be fully explored. To address this limitation, we propose the Uncertainty-Masked Bernoulli Diffusion (UMBD) model, the first generative refinement framework specifically designed for COD. UMBD introduces an uncertainty-guided masking mechanism that selectively applies Bernoulli diffusion to residual regions with poor segmentation quality, enabling targeted refinement while preserving correctly segmented areas. To support this process, we design the Hybrid Uncertainty Quantification Network (HUQNet), which employs a multi-branch architecture and fuses uncertainty from multiple sources to improve estimation accuracy. This enables adaptive guidance during the generative sampling process. The proposed UMBD framework can be seamlessly integrated with a wide range of existing Encoder-Decoder-based COD models, combining their discriminative capabilities with the generative advantages of diffusion-based refinement. Extensive experiments across multiple COD benchmarks demonstrate consistent performance improvements, achieving average gains of 5.5% in MAE and 3.2% in weighted F-measure with only modest computational overhead. Code will be released.

Problem

• Refining camouflaged object detection using generative diffusion
• Improving segmentation quality with uncertainty-guided masking
• Enhancing existing COD models with adaptive refinement

Innovation

• Uncertainty-guided masking for targeted refinement
• Hybrid Uncertainty Quantification Network (HUQNet)
• Seamless integration with Encoder-Decoder COD models