AI RESEARCH
SynMVCrowd: A Large Synthetic Benchmark for Multi-view Crowd Counting and Localization
arXiv CS.CV
•
ArXi:2603.23956v1 Announce Type: new Existing multi-view crowd counting and localization methods are evaluated under relatively small scenes with limited crowd numbers, camera views, and frames. This makes the evaluation and comparison of existing methods impractical, as small datasets are easily overfit by these methods. To avoid these issues, 3DROM proposes a data augmentation method. Instead, in this paper, we propose a large synthetic benchmark, SynMVCrowd, for practical evaluation and comparison of multi-view crowd counting and localization tasks.