On the Comparison of Classic and Deep Keypoint Detector and Descriptor Methods
This work provides a benchmark for computer vision researchers and practitioners to evaluate keypoint methods, but it is incremental as it compares existing methods without introducing new ones.
This study compared classic hand-crafted and deep learning-based keypoint detector and descriptor methods on the HPSequences dataset, finding that some classic combinations outperformed pretrained deep models in tasks like keypoint verification and image matching, with SuperPoint being the fastest in execution time.
The purpose of this study is to give a performance comparison between several classic hand-crafted and deep key-point detector and descriptor methods. In particular, we consider the following classical algorithms: SIFT, SURF, ORB, FAST, BRISK, MSER, HARRIS, KAZE, AKAZE, AGAST, GFTT, FREAK, BRIEF and RootSIFT, where a subset of all combinations is paired into detector-descriptor pipelines. Additionally, we analyze the performance of two recent and perspective deep detector-descriptor models, LF-Net and SuperPoint. Our benchmark relies on the HPSequences dataset that provides real and diverse images under various geometric and illumination changes. We analyze the performance on three evaluation tasks: keypoint verification, image matching and keypoint retrieval. The results show that certain classic and deep approaches are still comparable, with some classic detector-descriptor combinations overperforming pretrained deep models. In terms of the execution times of tested implementations, SuperPoint model is the fastest, followed by ORB. The source code is published on \url{https://github.com/kristijanbartol/keypoint-algorithms-benchmark}.