VisDrone2019-DET: Object Detection in Drone-Captured Images

A FiftyOne version of the VisDrone2019-DET dataset provides 8,629 samples of drone-captured imagery. It was published by Voxel51 and last updated on Hugging Face in May 2024. The dataset likely contains annotations for object detection tasks.

Use Cases

• Train object detection models based on drone-captured imagery
• Benchmark model performance on aerial perspective scenes
• Develop algorithms for analyzing traffic or crowd patterns from an overhead view

Strengths

• Contains 8,629 samples of drone imagery
• Published by Voxel51, a company specializing in dataset curation for computer vision

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

Voxel51

Freshness

Last updated 2024-05-03 19:07:50