Abstract:Concealed object detection (COD) in cluttered scenes is significant for various image processing applications. However, due to that concealed objects are always similar to their background, it is extremely hard to distinguish them. Here, the major obstacle is the tiny feature differences between the inside and outside object boundary region, which makes it trouble for existing COD methods to achieve accurate results. In this paper, considering that the surrounding environment information can be well utilized to identify the concealed objects, and thus, we propose a novel deep Surrounding-Aware Network, namely SurANet, for COD tasks, which introduces surrounding information into feature extraction and loss function to improve the discrimination. First, we enhance the semantics of feature maps using differential fusion of surrounding features to highlight concealed objects. Next, a Surrounding-Aware Contrastive Loss is applied to identify the concealed object via learning surrounding feature maps contrastively. Then, SurANet can be trained end-to-end with high efficiency via our proposed Spatial-Compressed Correlation Transmission strategy after our investigation of feature dynamics, and extensive experiments improve that such features can be well reserved respectively. Finally, experimental results demonstrate that the proposed SurANet outperforms state-of-the-art COD methods on multiple real datasets. Our source code will be available at <a class="link-external link-https" href="https://github.com/kyh433/SurANet" rel="external noopener nofollow">this https URL</a>.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in complex scenarios, the Concealed Object Detection (COD) task is difficult to distinguish because concealed objects are highly similar to the background. Specifically, the feature differences between the internal and external boundary areas of concealed objects are very small, making it difficult for existing COD methods to achieve accurate results. ### Core Challenges of the Problem 1. **Small Feature Differences**: Concealed objects are very similar to their surrounding environment in terms of texture, color, and shape, making them visually indistinguishable. 2. **Computational Efficiency**: How to distinguish these small feature differences at an acceptable computational cost while maintaining high precision. ### Overview of the Solution To solve the above - mentioned problems, the author proposes a new deep - learning network - SurANet (Surrounding - Aware Network). This network enhances feature extraction and the design of the loss function by introducing surrounding - environment information, thereby improving the recognition ability of concealed objects. Specifically: - **Introducing Surrounding - Environment Information**: By using surrounding - environment information to highlight the features of concealed objects, the differences between concealed objects and the background become more obvious. - **Efficient Contrastive Learning Strategy**: An efficient contrastive learning strategy is designed, enabling the network to better learn the differences between concealed objects and the surrounding environment during the training process. ### Main Contributions 1. **Discovering a New Perspective**: By focusing on surrounding - environment information, a new solution is provided to distinguish concealed objects. 2. **Proposing a New Network Architecture**: SurANet is designed, including the Surrounding - Aware Enhancement Module (SAE module) and the Surrounding - Aware Contrastive Loss (SAC loss function), which respectively enhance the feature representations between concealed objects and the surrounding environment. 3. **Proposing a New Training Strategy**: The Spatial - Compressed Correlation Transmission (SCCT) strategy is proposed. By optimizing the calculation method of the feature map, better performance is achieved without introducing significant additional computational costs. ### Summary This paper aims to solve the problems of small feature differences and computational efficiency in the concealed object detection task by introducing surrounding - environment information and an efficient contrastive learning strategy. Experimental results show that SurANet significantly outperforms existing COD methods on multiple real - world datasets.

SurANet: Surrounding-Aware Network for Concealed Object Detection via Highly-Efficient Interactive Contrastive Learning Strategy

Foreground Gating and Background Refining Network for Surveillance Object Detection

Go Closer to See Better: Camouflaged Object Detection via Object Area Amplification and Figure-Ground Conversion

Camouflaged Object Detection via Context-Aware Cross-Level Fusion

Attention guided multi-level feature aggregation network for camouflaged object detection

GLCONet: Learning Multi-source Perception Representation for Camouflaged Object Detection

Camouflaged object detection with counterfactual intervention

Salient Object Detection Based on Visual Perceptual Saturation and Two-Stream Hybrid Networks.

GLCONet: Learning Multisource Perception Representation for Camouflaged Object Detection

A Simple yet Effective Network based on Vision Transformer for Camouflaged Object and Salient Object Detection

I Can Find You! Boundary-Guided Separated Attention Network for Camouflaged Object Detection

SPANet: Spatial perceptual activation network for camouflaged object detection

MSCAF-Net: A General Framework for Camouflaged Object Detection via Learning Multi-Scale Context-Aware Features

CCNet: Collaborative Camouflaged Object Detection via decoder-induced information interaction and supervision refinement network

A Survey of Camouflaged Object Detection and Beyond

Feature-aware and iterative refinement network for camouflaged object detection

ZoomNeXt: A Unified Collaborative Pyramid Network for Camouflaged Object Detection

Camouflaged Object Detection with Feature Grafting and Distractor Aware

Detecting camouflaged objects via cross-level context supplement

Camouflaged Object Detection via Dual-branch Fusion and Dual Self-similarity constraints

FindNet: Can You Find Me? Boundary-and-Texture Enhancement Network for Camouflaged Object Detection