Neural network system, method and device used for vehicle part recognition

• Main Author: GUO, QING-PEI
• Format: Patent
• Language: Chinese; English
• Published: 16.03.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The embodiment of the invention provides a neural network system executed by a computer and used for vehicle part recognition, and the system comprises a convolution layer which is configured to receive a vehicle picture, carry out the convolution processing of the vehicle picture, and generate a convolution feature map corresponding to the vehicle picture; a region generation network RPN which isconfigured to determine a plurality of candidate regions taking vehicle components as potential targets based on the convolution feature map; a classification regression layer which is configured todetermine a component category and a frame corresponding to each candidate area based on the area characteristics of the candidate area; and a conditional random field CRF component which is configured to acquire the component category and the frame of each candidate area, extract the associated features of the plurality of candidate areas, and correct the component category and the frame of eachcandidate area based on the