Hybrid Improved Models Combined SR3 Module for Animal Recognition in Electric Car’s Actual Vision

With the development of information technology including artificial intelligence and image recognition technology, has gradually entered the application of electric cars. Especially real-time images monitoring in the field of electric car has become the eyes of them. In this paper, we proposed a hybrid improved intelligent recognition system based on SR3 module for animal recognition in the case of electric cars. The proposed hybrid improved system mainly used the actual image information collected by the monitor of the cars, which information acquired by the front-end monitors of electricity cars. And then the backend recognition model will train and run the classification algorithm to identify the animal species walk through so that the result can stop the vehicle by applying the brake order. The number of the test and train data set is about 2,000 animal images respectively, some of them were downloaded by Python crawler, they are all clear pictures under ideal conditions, and another 1000 pictures was taken by real cars monitor which are blurred picture under actual conditions. In terms of experimental results, several models of traditional and Deep Convolution Neural Networks by research and data analysis are compared, such as LBPH and VGG etc., under the real situation of the blur animal pictures which were taken by the real monitor, the accuracy of 88% of the hybrid improved model as a test set could be the best choice from several models. It could be the best electric cars' assistant visual model based on our findings.