An Effective Approach for Object Detection Using Deep Convolutional Networks

Computer vision and artificial intelligence has replaced many of the manual inspection systems in today’s world. In recent times, artificial intelligence is applied in developing of smart city services like object detection and tracking in surveillance system. However, these systems are impacted and yield poor performance due to two major reasons. The first reason is lower performance due to illumination changes or during poor illumination and the second problem is slow speed of operation. In this paper, the second problem of speed segmentation is addressed and the work focuses on developing a customized deep network that accelerate the process of object detection and evaluate the performance of the network with multiple video sequences. The approach is tested on standard datasets and compared against the state of art methods and observed to be more accurate and precise