Detection and Fast Tracking of Objects

Object detection refers to identifying a location where an object can be present. It also includes registering components of a particular object class at various levels of detail. For example, finding the faces in an image, finding the eyes and mouths of faces. One could require a precise outline of the object in the image or detection of well-defined landmarks on the object. Object Detection and Tracking in cluttered backgrounds is one of the most important problems in Computer Vision. Object Detection is receiving increased importance these days from computer vision researchers. The increased attention is motivated by its applications in Scene Understanding, Video Summarization, Content Based Retrieval, Surveillance, Manufacturing, inspection and world modeling. An Object Detection system should be able to detect objects in cluttered scenes even when they are partially occluded. In this project , we implement an object detector based on SIFT Features [1] and then use these features to track the object in the video. We also compared the performance of SIFT based object detector with the Haar Based Object Detector implemented in OpenCV [3]. The Haar Based Detector performed very badly because there was a single image of the book provided for training. Although we created a large dataset by applying transformations they did not capture all the possibilities. In the next section, we discuss our object detection approach followed by a discussion on approach used to track the book in Sec. 3. We then discuss the results in Sec. 4 before concluding in Sec.5.