Relocation method and device based on semantic segmentation and electronic equipment

The invention provides a relocation method and device based on semantic segmentation and electronic equipment, and the method comprises the steps: carrying out the semantic segmentation of a current environment image corresponding to the parking position of a current vehicle and a historically collected key environment image set, and obtaining a plurality of static object semantic categories corresponding to each image; determining a target environment image corresponding to the current environment image from the key environment image set by taking the topological relation of the semantic category of the static object as a descriptor of each image; the current vehicle is repositioned based on the position information corresponding to the target environment image, the problem that the image illumination robustness is poor is avoided, and the method can adapt to various vehicle application scenes. 本发明提供了一种基于语义分割的重定位方法、装置和电子设备，包括，对当前车辆所处停驻位置对应的当前环境图像和历史采集的关键环境图像集合分别进行语义分割，得到每张图像对应的多个静态对象语义类别；以静态对象语义类别的拓扑关系作为每张图像