UCL-based unmanned driving system object identification method and device

The invention provides a UCL-based unmanned driving system object identification method and device. The method comprises the following steps: firstly, carrying out automatic real-time target detection on a vehicle-mounted video by using a deep learning target detection model to extract object information; secondly, extending a unified content tag UCL, and newly adding an unmanned object attribute set UOPS; then, carrying out UCL semantic indexing on an object in the vehicle-mounted video according to the expanded unified content label; then, uploading the object identification indexed by the UCL to a knowledge base, and calibrating a UCL object information packet by using prior information in the knowledge base; and finally, reliably transmitting the UCL object information packet to a receiving module of an intelligent road network through a transmission module. According to the method, object data and processing software are separated and decoupled, the efficiency and the accuracy of vehicle-mounted video in