VMA: Divide-and-Conquer Vectorized Map Annotation System for Large-Scale Driving Scene.
Shaoyu ChenYunchi ZhangBencheng LiaoJiafeng XieTianheng ChengWei SuiQian ZhangChang HuangWenyu LiuXinggang WangPublished in: CoRR (2023)
Keyphrases
- d scene
- multiple images
- multimedia collections
- complex scenes
- video sequences
- active learning
- single image
- indoor and outdoor
- scene classification
- maximum a posteriori
- scene structure
- small scale
- three dimensional
- metadata
- line features
- outdoor scenes
- real scenes
- scene understanding
- real world
- image annotation
- image sequences
- image retrieval
- visual data
- automatic image annotation
- automatic annotation
- object recognition