APPLICATION OF DEEP LEARNING IN BIONIC EYE MONITORING SYSTEM

In the video monitoring system, due to the complicated background, environment light changes and restrictions on the performance of the device, resulting the difficulty in target detection algorithm design. While the traditional target detection algorithms typically rely on manual selection features, and hard to get a valid classification from amounts of data. Based on the deep learning algorithm, this paper constructs a convolutional neural network, and using the person, vehicle graphics, which are collected by the bionic eye monitoring system, for training the network. Besides we design a series of experiments for analysis the characteristics of the network, and proved that the training set distribution of samples in each category and the number of training samples in a small case would impact the results of the training of deep learning network. In addition, it shows that the parameters of the model using the fuzzy image for training can get a similar result with fuzzy image and clear image.