Visual earth surface attribute classification prediction model training method and system

• Main Authors: LI HONGZE, XIE RUI, ZHOU HAOTIAN, ZHAO HUIJING
• Format: Patent
• Language: Chinese; English
• Published: 02.04.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention provides a visual earth surface attribute classification prediction model training method and system, and belongs to the technical field of automatic driving. Based on the training data, a visual earth surface attribute classification prediction model is trained, and the visual earth surface attribute classification prediction model comprises a visual module, an organism perception module and a classification network; the input of the visual module is one frame of image block, and the output is an earth surface attribute label; the input of the machine body sensing module is a time sequence signal of a 100-millisecond accelerometer and a wheel speed encoder, and the output of the machine body sensing module is an earth surface attribute label; and the classification network learns model parameters of the visual module and the body sensing module in a full supervision mode based on a manually labeled sample to obtain earth surface attribute classification. According to the method, the characteris