The Object Detection Algorithm Research And Application Based On Deep Learning

Object detection is a challenging problem in the field of computer vision and which main purpose is to detect and locate specific goals from static images or video. It is based on the technology of technology of image processing, pattern recognition,artificial intelligence and automatic control and widely used in traffic accident prevention, suspicious warned of dangerous goods in factory, military restricted zone monitoring and senior human-computer interaction. The current lack of a mature and general method to detect object because of the environment is complicated. Object detection research exist opportunities and challenges in practical application.This thesis first analyzes the domestic and foreign research status of object detection algorithm, emphatically introduces the application method which are widely used is based on the object feature trained classifier to classify object. Because of the existing feature of the trained classifier to classify object has high false positives rate,this thesis present a pedestrian object detection algorithm based on convolution neural network on the basis of deep learning. The algorithm consists of two steps in order to solve the low efficiency of sliding window with convolution neural network,(1) the suspected pedestrian window confirmation;(2) the pedestrian detection. In suspected existing pedestrian window confirmation, this thesis use the fusion feature as the description of the pedestrian training classifier and the ideal of nearby scale feature similar to build classifier pyramid. On the inspected images, this thesis use different scales of sliding window to slide traversal to confirm suspected exist pedestrian window.In the pedestrian detection, this thesis rely a large number of positive and negative samples to train and get a convolution neural network. In order to better adept the pedestrian detection, this thesis improve the topology of traditional convolution network.Input suspected existence of pedestrian’s window into the improved convolution neural network to detect the pedestrian.In order to verify the accuracy of the proposed algorithm, this thesis test pedestrian detection experiments in the INRIA pedestrian database. Separately treat each window and each image as detection unit, this thesis statistics the detection rate and error detection rate of the algorithm. On the standard of the existence of an error in every image detection window, this thesis gets 93% detection rate. Compared theexperimental results with train detector using ACF feature, under the same false positive rate, the algorithm in this thesis has 3% detection rate higher than the detector trained from ACF feature and detection time less four folds than single use convolutional neural networks. The experimental results certify the effectiveness of the algorithm in this thesis.