Plug-and-Play DISep: Separating Dense Instances for Scene-to-Pixel Weakly-Supervised Change Detection in High-Resolution Remote Sensing Images

TL;DR

This paper introduces DISep, a plug-and-play method that separates dense instances in weakly-supervised change detection, significantly improving accuracy in high-resolution remote sensing images by addressing the problem of instance lumping.

Contribution

We propose a novel Dense Instance Separation (DISep) method that refines pixel features for better instance separation under scene-level supervision, enhancing existing change detection models.

Findings

Achieved state-of-the-art performance on multiple datasets.

Enhanced both Transformer-based and ConvNet-based methods.

Minimal additional training cost with no inference overhead.

Abstract

Existing Weakly-Supervised Change Detection (WSCD) methods often encounter the problem of "instance lumping" under scene-level supervision, particularly in scenarios with a dense distribution of changed instances (i.e., changed objects). In these scenarios, unchanged pixels between changed instances are also mistakenly identified as changed, causing multiple changes to be mistakenly viewed as one. In practical applications, this issue prevents the accurate quantification of the number of changes. To address this issue, we propose a Dense Instance Separation (DISep) method as a plug-and-play solution, refining pixel features from a unified instance perspective under scene-level supervision. Specifically, our DISep comprises a three-step iterative training process: 1) Instance Localization: We locate instance candidate regions for changed pixels using high-pass class activation maps. 2) Instance Retrieval: We identify and group these changed pixels into different instance IDs through connectivity searching. Then, based on the assigned instance IDs, we extract corresponding pixel-level features on a per-instance basis. 3) Instance Separation: We introduce a separation loss to enforce intra-instance pixel consistency in the embedding space, thereby ensuring separable instance feature representations. The proposed DISep adds only minimal training cost and no inference cost. It can be seamlessly integrated to enhance existing WSCD methods. We achieve state-of-the-art performance by enhancing {three Transformer-based and four ConvNet-based methods} on the LEVIR-CD, WHU-CD, DSIFN-CD, SYSU-CD, and CDD datasets. Additionally, our DISep can be used to improve fully-supervised change detection methods. Code is available at https://github.com/zhenghuizhao/Plug-and-Play-DISep-for-Change-Detection.