Image
Visualization: Images are displayed in the image viewer with pan, zoom, and pixel-level inspection. When part of an MCAP recording, images appear in dedicated Image panels synchronized with other sensor streams. Annotation: Each image is annotated independently as a single frame. All seven 2D annotation tools are available: bounding boxes, polygons, segmentation masks, polylines, keypoints, classification, and object tracking (in video sequences). Supported formats: JPEG, PNG, WebP, BMP Use cases: Object detection, instance segmentation, semantic segmentation, image classification, keypoint detection, pose estimation.Video
Visualization: Videos are automatically converted to frame sequences on upload, enabling frame-by-frame playback with timeline scrubbing. Navigate forward and backward through frames, jump to specific timestamps, or play back at configurable speeds. Annotation: Annotators work frame-by-frame with object tracking across the timeline. Object IDs persist across frames for consistent identity assignment. All 2D annotation tools are available on each frame. Supported formats: MP4, MOV Use cases: Object tracking, action recognition, temporal event detection, driving scene labeling, behavior analysis.Video processing happens in the background after upload. Large videos may take several minutes to convert. You can monitor sequence status in Mission Control or via the API.
LiDAR / Point Cloud
Visualization: Point clouds are rendered in a 3D viewer with bird’s-eye view, perspective view, and side views. Six visualization modes let you color points by different properties:| Mode | Description |
|---|---|
| Neutral | Single uniform color for structural overview |
| Intensity | Return strength — highlights reflective surfaces |
| Rainbow | Temporal or sequential coloring |
| Label | Semantic class coloring from annotations |
| Panoptic | Instance-level coloring for individual objects |
| Image Projection | Camera imagery projected onto the point cloud |
MCAP
Visualization: MCAP recordings are displayed in the multi-sensor viewer with up to eight panel types: Image, 3D / Point Cloud, Plot, Raw Messages, Log, Map, Gauge, and State Transitions. Avala automatically detects message types in the recording and assigns topics to the appropriate panel type. All panels share a synchronized timeline for coordinated playback of camera, LiDAR, radar, IMU, and other sensor streams. The layout composer automatically builds an optimized panel arrangement based on the topics in your recording, or you can customize the layout manually. Navigate frame-by-frame, scrub to any timestamp, or play back the full recording with configurable speed. Annotation: Avala parses MCAP files to extract and synchronize sensor streams. Camera images are displayed alongside projected LiDAR data, enabling multi-camera annotation with 3D context. Annotators place 3D cuboids that project consistently across all camera views. Supported formats: MCAP (with ROS message support) Use cases: Multi-sensor fusion, surround-view perception, autonomous vehicle data labeling, robotics sensor calibration, fleet data review.MCAP support includes automatic extraction of camera intrinsics and extrinsics for accurate LiDAR-to-camera projection. Both pinhole and double-sphere (fisheye) camera models are supported. See the MCAP / ROS integration guide for setup details.
Splat
Visualization: Gaussian Splat scenes are rendered in a WebGPU-accelerated 3D viewer. Navigate freely through photorealistic 3D scene reconstructions with smooth camera controls. The renderer uses GPU radix sorting, buffer pooling, and pipeline precompilation for real-time performance. Annotation: Annotators navigate the reconstructed environment and place 3D annotations directly in the scene. Classification labels can be applied to the full scene or individual regions. Supported formats: Gaussian Splat Use cases: 3D scene understanding, novel view synthesis annotation, spatial AI training data, environment mapping.Capabilities Comparison
The following table shows visualization and annotation capabilities for each data type:| Capability | Image | Video | Point Cloud | MCAP | Splat |
|---|---|---|---|---|---|
| Visualization | |||||
| 2D Image Viewer | Yes | Yes | — | Yes | — |
| 3D Point Cloud Viewer | — | — | Yes | Yes | — |
| 3D Splat Viewer | — | — | — | — | Yes |
| Multi-Panel Layout | — | — | — | Yes | — |
| Timeline Playback | — | Yes | Yes | Yes | — |
| Visualization Modes (6) | — | — | Yes | Yes | — |
| Annotation | |||||
| Bounding Box | Yes | Yes | — | — | — |
| Polygon | Yes | Yes | — | — | — |
| 3D Cuboid | — | — | Yes | Yes | Yes |
| Segmentation | Yes | Yes | — | — | — |
| Polyline | Yes | Yes | — | — | — |
| Keypoints | Yes | Yes | — | — | — |
| Classification | Yes | Yes | Yes | Yes | Yes |
| Object Tracking | — | Yes | Yes | Yes | — |
Upload Requirements
| Property | Limit |
|---|---|
| Max file size (images) | 20 MB per file |
| Max file size (video) | 2 GB per file |
| Max file size (point cloud) | 500 MB per file |
| Max file size (MCAP) | 5 GB per file |
| Supported image formats | JPEG, PNG, WebP, BMP |
| Supported video formats | MP4, MOV |
| Supported point cloud formats | PCD, PLY |
| Supported multi-sensor formats | MCAP |
Next Steps
Managing Datasets
Upload, organize, and manage your data in Mission Control.
MCAP / ROS Integration
Set up multi-sensor data pipelines with MCAP and ROS.
Core Concepts
Understand viewers, panels, layouts, and other platform concepts.
Architecture
Learn how the visualization engine and backend services work together.