PROJECT TITLE :

Convolutional Neural Network Based Automatic Object Detection on Aerial Images

ABSTRACT:

We are witnessing daily acquisition of enormous amounts of aerial and satellite imagery. Analysis of such giant quantities of data can be useful for several practical applications. In this letter, we tend to present an automatic content-based analysis of aerial imagery so as to detect and mark arbitrary objects or regions in high-resolution pictures. For that purpose, we tend to proposed a method for automatic object detection based on a convolutional neural network. A novel 2-stage approach for network training is implemented and verified within the tasks of aerial image classification and object detection. Initial, we tested the proposed coaching approach using UCMerced data set of aerial images and achieved accuracy of approximately ninety eight.half-dozen%. Second, the strategy for automatic object detection was implemented and verified. For implementation on GPGPU, a required processing time for one aerial image of size 5000 5000 pixels was around thirty s.