Tianyang Dong

Remote Sensing (RS) Change Detection (CD), a technique focused on identifying surface alterations from bi-temporal images, holds substantial significance for various applications, such as land management and disaster surveillance. In the last decade, deep learning-based CD methods have advanced rapidly. However, recognizing sporadic distributional changes over time in complex scenes remains a significant challenge. However, many existing solutions rely on large model capacities and high computational costs, yet still fail to incorporate sufficient semantic information for accurately recognizing complex real changes. To tackle this challenge, a lightweight full-information and dual-guide network (Lighter) for RSCD is presented. Specifically, we design a lightweight full-information mingling module (FIMM) that emphasizes the injection of multi-perspective information during the feature interaction. This approach leverages rich semantics as cues to reason about diverse changes. Furthermore, we propose a lightweight dual-guide difference capture module (DDCM), which utilizes the unique information of each guide to teach each other thereby reducing the interference of pseudo-variations. Extensive experiments on four datasets demonstrate that our lightweight architecture achieves state-of-the-art (SOTA) performance with only 1.10M parameters and 2.01G FLOPs. The code will be released upon publication.

Change detection (CD) in optical remote sensing images has advanced significantly with the adoption of deep learning. However, CD inherently faces two challenges: (1) varying sizes and shapes of change regions arising from spatial complexity, and (2) pseudo-changes caused by temporal migrations, such as seasonal variations and lighting differences. Existing methods typically focus on single-grained information during the feature interaction stage, which limits their ability to perceive changes in features of varying sizes. Additionally, these methods inadequately address ambiguous regions during the difference capture stage, making them highly susceptible to pseudo-change interference. To address these challenges, we propose a novel network called the Spatio-temporal Multi-granularity Intermingling Network (STMINet). First, we introduce the Spatio-temporal Multi-granularity Interleaving Module (STMIM) to capture multi-granularity information across both time and space, greatly enhancing the detection of changes in features of varying sizes and shapes. Second, we propose the Multi-branch Differential Acquisition (MDA), which incorporates information from inconspicuous regions and mitigates pseudo-change interference through a three-branch design. Experimental results on four publicly available datasets (LEVIR-CD, GZ-CD, WHU-CD, and SYSU-CD) demonstrate that STMINet significantly outperforms state-of-the-art methods in performance metrics and visualization, achieving F1 score improvements of 0.22–3.42% over existing approaches while employing a simple ResNet-18 backbone. The source code will be published at https://github.com/heikeyuhuajia/STMINet.

In complex stage scenes, the accuracy of actor detection is severely affected by color bias and uneven illumination problems generated by multiple light sources. To solve these problems, an actor detection method based on pseudo-multimodal fusion is proposed. First, a random illumination method is used to construct an enhanced image to form a pseudo-multimodal image pair with the original image; second, an actor key point candidate set is created in the enhanced image, and some key points are randomly selected from the set to construct enhanced patches, and the patches are replaced into the original image for training. Then, based on the traditional feature pyramid network, the visual attention module is used to construct visual attention encoders to enhance the interaction logic of multi-scale features. On a self-constructed dataset of 4543 images which containing stage actors, the average accuracy of detection is improved 0.4% to 2.9% by combining the three models, and the results show that the proposed method can better reduce the effects of color bias and uneven illumination.