Semi-supervised semantic segmentation with multi-reliability and multi-level feature augmentation

Jianjian Yin, Zhichao Zheng, Yulu Pan, Yanhui Gu, Yi Chen

Abstract: Highlights•Addressing information loss through multi-reliability feature augmentation.•Resolving the uneven number of category pixels by multi-level feature augmentation.•Various data perturbations enhance the robustness of the model.•Experiments showing that our method is comparable to advanced methods.