Object Detection and Classification Using GPU Acceleration

In order to speed up the image processing for self-driving cars, we propose a solution for fast vehicle classification using GPU computation. Our solution uses Histogram of Oriented Gradients (HOG) for feature extraction and Support Vector Machines (SVM) for classification. Our algorithm achieves a higher processing rate in frames per second (FPS) by using multi-core GPUs without compromising on its accuracy. The implementation of our GPU programming is in OpenCL, which is a platform independent library. We used a dataset of images of cars and other non-car objects on road to feed it to the classifier.