Semantic segmentation is one of the earliest research directions in the field of artificial intelligence. With the wave of AI algorithms landing, how to effectively counter complex and harsh backgrounds in real scenarios, and how to achieve real-time performance on mobile terminals with limited computing capacity come to be the latest trend. In detail, for foggy, heavy rain, underwater, cloudy and other environments, where the illumination is weak, the segmentation object is blocked, and the edges are not obvious, some targeted lightweight network structures are badly needed to solve these problems. This special issue aims to collect the latest research in this field and realize the transformation from academia to industry. Interested topics include but are not limited to real time semantic segmentation algorithms, noise (deblurring, deraining, dehazing and so on) reduction algorithms, algorithms that combine noise reduction and semantic segmentation, and the application of semantic segmentation algorithms on mobile terminals.