The CNN and DPM based approach for multiple object detection in images

摘要

With the development of intelligent device and social media, the bulk of data on Internet has grown with high speed. There are so many important aspect in image processing, object detection is one of the international demanded research field. Multiple object detection is an important concept in object detection. In object detection extracting the features and handling the occlusion are two most important components. A Region-based Convolution Neural Network (R-CNN) has achieved great success in extracting the region based features which may used for extracting multiple regions from the images and Deformable Part Based Model (DPM) improve the ability for handling the occlusion. Occlusion handling is nothing but when multiple objects are near to each other that time some objects are not detected so this drawback will be handled by DPM. Existing method not performing well in the aspect of detecting multiple objects. In this paper R-CNN and DPM are to be integrated to detect multiple objects. By combining these two models we are able to notice every single object with high accuracy.