Total: 1
Bag-of-words model is implemented and tried on 10-class visual concept detection problem. The experimental results show that "DURF+ERT+SVM" outperforms "SIFT+ERT+SVM" both in detection performance and computation efficiency. Besides, combining DURF and SIFT results in even better detection performance. Real-time object detection using SIFT and RANSAC is also tried on simple objects, e.g. drink can, and good result is achieved.