Semantic segmentation model training method and device, semantic segmentation method and device, equipment and medium

The embodiment of the invention discloses a semantic segmentation model training method and device, a segmentation method and device, equipment and a medium, and the method comprises the steps: obtaining sample input data which comprises sample road image information collected by image collection equipment; inputting the sample input data into a preset semantic segmentation model for processing, the preset semantic segmentation model being used for extracting spatial feature information and semantic feature information of the sample input data, fusing the spatial feature information and the semantic feature information, and determining a segmentation result of road semantics according to a fusion result; and determining a target loss relationship according to a segmentation result output by the preset semantic segmentation model, and training the preset semantic segmentation model according to the target loss relationship. According to the method, the road semantic segmentation result is determined according