Target detection and recognition for wide area border defense intelligent surveillance

A lot of exploration and practice on the construction of border video surveillance under the conditions of informatization have been investigated. However, there are still some problems in the video surveillance system, such as insufficient application depth, insufficient information analysis and processing capabilities, and low level of intelligent application. With the rapid development of deep learning and computer vision technology, intelligent video surveillance systems are widely used in various fields such as transportation and security. How to apply intelligent video surveillance to border defense systems to meet the requirements of practical, effective, reliable, advanced and econom-ical. In response to the above needs, based on the three-camera array video surveillance image, this paper sets five types of monitoring targets for people, car, airplane, rotorcraft and birds, and studies the application of MobileNet-SSD network for target detection and recognition under the Caffe framework. The paper firstly choose the Caffe framework, MobileNet-SSD network and OpenCV to build a working environment. Secondly, build a specific dataset based on five types of images with established targets and VOC data sets. Finally, image and video target detection tests are carried out on the model generated by training.