A lightweight image segmentation algorithm based on MobileNetv3 and global context block attention mechanism

Image segmentation is a crucial task in computer vision that involves dividing an image into distinct regions, with each region containing pixels that share similar attributes. Traditional methods like Otsu, Sobel, and Canny often suffer from high computational complexity and sensitivity to parameters. To address these limitations, this paper proposes a lightweight image segmentation algorithm that integrates MobileNetv3-Small with a global context block attention mechanism. MobileNetv3 leverages depthwise separable convolutions and optimized architecture to significantly reduce computational load and parameter count. To further improve segmentation accuracy, a deformable convolutional network is employed during feature extraction, while the global context block attention mechanism enhances focus on target regions. The proposed algorithm not only improves segmentation performance but also offers a lightweight solution suitable for resource-constrained environments.