Image semantic segmentation is a key technology in computer vision, widely used in scenarios such as autonomous driving, intelligent security, and medical image analysis. However, existing semantic segmentation models often face a contradiction between large number of parameters, high computational complexity, and insufficient segmentation accuracy. To address this issue, this paper designs and implements a lightweight image semantic segmentation model incorporating an attention mechanism. This model is based on an improved DeepLabV3+ framework, using MobileNetV2 as a lightweight backbone network. A Hybrid Domain Attention Module (CBAM) is introduced in the encoding stage to enhance the representation ability of important features. Simultaneously, a Channel Hollow Spatial Pyramid Pooling Module (C-ASPP) is designed to weight each channel while extracting multi-scale contextual information. In the decoding stage, a dense neighborhood prediction module is introduced to fuse high- and low-level features, refining the segmentation boundaries. Experimental results on the PASCAL VOC 2012 and Cityscapes datasets show that the proposed model achieves an average intersection-over-union (mIoU) of 74.83% and 75.21% respectively with only 24.3 MB of parameters, representing improvements of 3.21% and 2.94% over the baseline model, achieving a good balance between segmentation accuracy and computational efficiency.

Xia Jiayue, Long Yanbin. Design and Implementation of a Lightweight Image Semantic Segmentation Model Combining Attention Mechanism. International Journal of New Developments in Engineering and Society (2026), Vol. 10, Issue 1: 57-63. https://doi.org/10.25236/IJNDES.2026.100108.

[1] Ma Dongmei, Wang Pengyu, Guo Zhihao. A lightweight semantic segmentation based on attention mechanism[J]. Computer Engineering and Science, 2024, 46(8): 1503-1512.

[2] Yuan Manman, Lu Hao. Semantic segmentation algorithm integrating lightweight and attention mechanism[J]. Journal of Chongqing Technology and Business University (Natural Science Edition), 2025(1): 57-63.

[3] Zhou Hanlian, Ye Qing, Liu Wenqi. A lightweight improved UNet small target aircraft cross-domain segmentation model based on domain adaptation[J]. Optoelectronic Engineering, 2026. 

[4] Ma D, Wang P, Guo Z. An improved lightweight semantic segmentation network with dual attention[J]. Journal of Computer Engineering and Science, 2024, 46(9): 1620-1628.

[5] Wang Guogang, Dong Zhihao. Lightweight image semantic segmentation based on attention mechanism and dense neighborhood prediction[J]. Computer Science, 2024, 51(6A): 230300204.

[6] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.

[7] Woo S, Park J, Lee J Y, et al. CBAM: Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision, 2018: 3-19.

[8] Chen L C, Zhu Y, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of the European Conference on Computer Vision, 2018: 801-818.

[9] Sandler M, Howard A, Zhu M, et al. MobileNetV2: Inverted residuals and linear bottlenecks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 4510-4520.

[10] Yu C, Wang J, Peng C, et al. BiSeNet: Bilateral segmentation network for real-time semantic segmentation[C]//Proceedings of the European Conference on Computer Vision, 2018: 325-341.