ApolloScape: 3D LiDAR and Car Instance Data for Autonomous Driving

ApolloScapeAuto provides this collection of autonomous driving data featuring 3D LiDAR point clouds and 3D car instance annotations. The repository, last updated in May 2024, serves as a resource for training models in trajectory prediction and video inpainting. It includes specialized labels for complex spatial and temporal tasks in the automotive domain.

Use Cases

• Trajectory prediction using historical movement sequences
• 3D object detection using 3D LiDAR point cloud data
• Scene reconstruction and object removal using video-inpainting labels

Strengths

• Includes 3D LiDAR point clouds for spatial analysis
• Features 3D car instance labels for precise object modeling
• Provides video sequences for temporal task evaluation

Limitations

• Metadata does not specify the total number of frames or records
• Sensor calibration details and environmental conditions are not provided in the summary

Provenance

Source

ApolloScapeAuto

Collection Method

Sensor-based collection including LiDAR and cameras

Freshness

Last updated May 2024