SeqNet: An Interest Point Detector Based on Unsupervised Learning
Jing‐Yi Lin, Xiaorui Hu, Dong Li, Yun Zhang · 2019
To obtain robust, stable and well-defined image features, an excellent interest point detector is necessary. Interest points in images refer to some special points that are useful for subsequent processing tasks, such as edges, comers, textures, blobs, etc. Since the task is important in computer vision, more and more learning-based methods have been proposed. However, most of their methods are built upon hand-crafted methods, and the performance of hand-crafted detectors limits their further development. In this paper, we propose SeqNet, an unsupervised learning network for detecting interest points. We qualitatively and quantitatively show that our method performs better or on-par with baselines.