04-04-2023 дата публикации
Номер: CN115908813A
Принадлежит:
The invention belongs to the technical field of scene segmentation and automatic driving, and discloses a real-time semantic segmentation model and method in a road scene, and the model comprises an initialization module, a channel attention module, a down-sampling module, a mixed hole grouping module, and an up-sampling module. According to the technical scheme of the invention, a mixed cavity grouping module is constructed by using decomposition convolution, depth separable convolution and cavity convolution, local and context information is extracted in a simple and effective manner, then information interaction between channels is captured by using a channel attention module, and finally, information interaction between channels is realized. Characteristic branches from different stages in the network are subjected to feature fusion in a layer-skipping connection mode, so that shallow features and deep advanced semantic information are fused, feature representation is enhanced, segmentation ...
Подробнее