Scholarlink

Video surveillance is one of the most essential tasks in the real world. Therefore, many state-of-the-art tracking algorithms were developed that can be applied for general video surveillance purposes. However, due to lack of dedicated benchmark for pedestrian tracking algorithms, this domain is least explored. Further, the commercial availability and low cost make unmanned aerial vehicles (UAVs) more suitable for tracking from the sky. Therefore, the present work proposes a new benchmark using aerial datasets for visual pedestrian tracking by UAVs. This newly developed p-benchmark consists of 117 real-world image sequences to exhaustively evaluate the performance of different Siamese network-based trackers, which may act as a baseline.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

P-Benchmark: A Benchmark for Video Surveillance Using UAVs

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Himanshu Gupta,
  • Deepak Jangid,
  • Sourabh Verma,
  • Om Prakash Verma,
  • Tarun K. Sharma

摘要

Video surveillance is one of the most essential tasks in the real world. Therefore, many state-of-the-art tracking algorithms were developed that can be applied for general video surveillance purposes. However, due to lack of dedicated benchmark for pedestrian tracking algorithms, this domain is least explored. Further, the commercial availability and low cost make unmanned aerial vehicles (UAVs) more suitable for tracking from the sky. Therefore, the present work proposes a new benchmark using aerial datasets for visual pedestrian tracking by UAVs. This newly developed p-benchmark consists of 117 real-world image sequences to exhaustively evaluate the performance of different Siamese network-based trackers, which may act as a baseline.