Three dynamic basketball court scenes captured in Guangzhou, Dongguan, and Long360. Each scene contains 36 viewpoints with videos at 1080p or 4K resolution, and the Guangzhou and Dongguan sequences consist of approximately 250 frames each. The dataset was created by BestWJH and is intended for evaluating dynamic scene reconstruction methods.
Use Cases
- Evaluate dynamic scene reconstruction algorithms using multi-view video sequences from 36 viewpoints.
- Analyze large-scale motion patterns across three distinct basketball court scenes.
- Benchmark 3D reconstruction methods on 1080p and 4K video frames from synchronized viewpoints.
Strengths
- Three distinct dynamic scenes provide varied test environments.
- Each scene offers 36 synchronized video viewpoints for comprehensive multi-view analysis.
- Video sequences are captured at high resolution (1080p or 4K).
Limitations
- Limited temporal scope with sequences approximately 10 seconds long.
- Geographic coverage is restricted to three locations in China.
- Unknown total data size, row count, and specific file formats.
Provenance
- Source
- huggingface
- Collection Method
- Multi-view video capture from 36 viewpoints per scene.
- Time Range
- null
- Freshness
- Last updated March 2026.
- Geography
- Guangzhou (gz), Dongguan (dg), and Long360, China.