Object classification with convolutional neural networks on embedded systems

Deep Neural Network is successfully applied for image classification problems. The understanding of what object is detected in the image is great interested. The prupose of this article is that image classification problem and tracking classified object could be applied in real time systems. The aim of this thesis is image classification process is observing how successful in an embedded system. In this work, proposed architecture has only 3 GB memory with accuracy %85.9. It trains in 4 hour,and outputs the classification result in 0.7 ms.