Search

Due to the severe distortion of the fisheye image, the rectangular bounding box contain a lot of invalid information. So the multi-point representation method are emerging. However, it will fail in some extreme cases especially when the center of the object is not in the instance. In this work, we propose a Convex Hull Regression Strategy for people detection on top-view fisheye images. It replaces the instance with its convex hull to solve the above challenging issue and can be pre-trained on regular datasets without additional processing. In addition, the mosaic and mixup data augmentation methods that perform well under rectangular boxes are applied to our representation. Finally, we improve the label assignment and propose a more reasonable loss function called PDIoU loss so as to focus on the overall IoU between ground truth polygon and predicted polygon. Experimental results demonstrate that our method outperforms state-of-the-art algorithms. Source code is available at https://github.com/xiaoxuebajie/CHRS. [ABSTRACT FROM AUTHOR]

The UAVs captured targets are relatively small when compared with the ordinary surveillance cameras. Thus, a strong discriminative ability is required for aerial trackers to accurately locate small targets, especially in challenging scenes, including similar objects, occlusion and size change. Since most existing aerial trackers do not perform satisfactorily, in this paper, we design a tracker called TAP-GC that utilizes a weight-sharing deep CNN network to extract the multi-scale template and test features. We then construct a target-aware pooling module in the template branch, allowing the tracker to pay more attention to the target-related information. Thereafter, we directly fuse the template and test features through a transformer which is able to make full use of the global context, enabling the tracker to discriminate the target more accurately. Extensive experiments on well-known aerial tracking benchmarks, UAV123, UAV123@10fps and DTB70, show that our tracker outperforms a number of state-of-the-art trackers. In addition, when evaluating TAP-GC on OTB100, a tracking benchmark captured by ordinary cameras, it also achieves leading tracking performance. TAP-GC can achieve about 70 fps speed for real-time UAV tracking. [ABSTRACT FROM AUTHOR]