Vehicle Type Recognition Algorithm Based on Improved Network in Network

Vehicle type recognition algorithms are broadly used in intelligent transportation, but the accuracy of cannot meet requirements production application. For high efficiency multilayer perceptive layer Network (NIN), nonlinear features local receptive field images can be extracted. Global average pooling (GAP) avoid network from overfitting, and small convolution kernel decrease dimensionality feature map, as well downregulate number model training parameters. On that basis, residual error is adopted to build a novel NIN by altering size layout original NIN. The feasibility algorithm verified based on Stanford Cars dataset. By properly setting weights learning rates, for vehicle reaches 97.2%.