LHDACT: Lightweight Hybrid Dual Attention CNN and Transformer Network for Remote Sensing Image Change Detection

With the significant advancements of deep learning (DL) in the field of remote sensing (RS) imagery, a plethora of change detection (CD) methods based on CNNs, attention mechanisms, and transformers (TRs) have emerged. Presently, a substantial amount of research has gradually relinquished control over parameter quantities in pursuit of enhanced outcomes, resulting in the inflation of networks with numerous stacked modules. This letter is dedicated to integrating lightweight approaches into the CD task. We introduce a lightweight hybrid dual-attention CNN and TR (LHDACT) network based on depthwise over-parameterized convolutional (DO-Conv). In comparison to traditional convolution, DO-Conv combines both traditional and depthwise convolutions, achieving commendable performance enhancement with minimal additional cost. Furthermore, we leverage DO-Conv to enhance the multiscale average pooling (MSAP) module, ensuring global context with low computational overhead. To better discern regions of interest within complex images, we enhance the dual attention module (DAM) by sharing weights across spatial and channel dimensions, thereby bolstering feature region identification. At last, we employ a compact TR module to capture feature differences, enabling precise CD. Our approach is evaluated on the LEVIR-CD, WHU-CD, and GZ-CD datasets, yielding F1 scores of 91.23%, 87.51%, and 85.32%, respectively. These results demonstrate high performance on a cost-effective scale.