Retrospective Memory for Camouflaged Object Detection

TL;DR

This paper introduces RetroMem, a novel recall-augmented architecture for camouflaged object detection that integrates historical context through memory mechanisms, significantly enhancing detection accuracy in complex scenes.

Contribution

The paper proposes RetroMem, a two-stage training framework with a dense multi-scale adapter and dynamic memory, enabling better camouflage pattern understanding and improved detection performance.

Findings

RetroMem outperforms existing state-of-the-art methods on multiple datasets.

The dense multi-scale adapter enhances feature extraction with minimal parameters.

Memory mechanisms effectively leverage historical knowledge for better inference.

Abstract

Camouflaged object detection (COD) primarily focuses on learning subtle yet discriminative representations from complex scenes. Existing methods predominantly follow the parametric feedforward architecture based on static visual representation modeling. However, they lack explicit mechanisms for acquiring historical context, limiting their adaptation and effectiveness in handling challenging camouflage scenes. In this paper, we propose a recall-augmented COD architecture, namely RetroMem, which dynamically modulates camouflage pattern perception and inference by integrating relevant historical knowledge into the process. Specifically, RetroMem employs a two-stage training paradigm consisting of a learning stage and a recall stage to construct, update, and utilize memory representations effectively. During the learning stage, we design a dense multi-scale adapter (DMA) to improve the pretrained encoder's capability to capture rich multi-scale visual information with very few trainable parameters, thereby providing foundational inferences. In the recall stage, we propose a dynamic memory mechanism (DMM) and an inference pattern reconstruction (IPR). These components fully leverage the latent relationships between learned knowledge and current sample context to reconstruct the inference of camouflage patterns, thereby significantly improving the model's understanding of camouflage scenes. Extensive experiments on several widely used datasets demonstrate that our RetroMem significantly outperforms existing state-of-the-art methods.