Dark Mode
Asia Stereo Egocentric Vision Dataset with Calibration
Egocentric & First-Person Data
Tags and Keywords
£162,500
About
The Asia Stereo Egocentric Vision Dataset is a multimodal first-person stereo video dataset collected across industrial (60%) and residential (40%) environments in Asia using a GoPro or equivalent camera system. The dataset contains synchronized left and right camera streams recorded in 4K and 1080p at 30/60 FPS with H.264/H.265 encoding and includes audio. Although native depth maps are not provided, the dataset includes full stereo calibration parameters, enabling accurate depth and disparity estimation using standard stereo matching algorithms. The dataset is designed for robotics, computer vision, embodied AI, stereo perception, visual SLAM, and autonomous navigation research.
Data Product Features
| Feature | Description |
|---|---|
| Environment | 60% Industrial, 40% Residential |
| Data Type | Stereo Egocentric Video |
| Region | Asia |
| Camera | GoPro or equivalent |
| Viewpoint | First-Person Stereo |
| Stereo Calibration | Included |
| Native Depth Maps | No |
| Resolution | 4K / 1080p |
| Frame Rate | 30 FPS / 60 FPS |
| Video Codec | H.264 / H.265 |
| Audio | Yes |
| Motion Level | Low to Medium |
Distribution
Format
- MP4 video files
- Stereo calibration files (camera parameters)
Encoding
- H.264 / H.265
Structure
- Folder-based organization containing synchronized left and right camera recordings with calibration files.
Data Volume
- Approximate Duration: 50 hours
- Approximate Storage Size: 350 GB
- Approximate Records: 5,000 synchronized stereo recordings
Usage
This dataset is ideal for a variety of applications:
- Stereo Vision: Depth and disparity estimation using calibrated stereo pairs.
- Visual SLAM: Simultaneous localization and mapping.
- Robot Navigation: Localization and obstacle perception.
- Embodied AI: First-person perception and decision-making.
- Computer Vision: Scene understanding, object detection, and segmentation.
- 3D Reconstruction: Multi-view scene reconstruction.
- Autonomous Systems: Perception for robots and autonomous platforms.
- Academic Research: Robotics and computer vision research.
Coverage
Geographic Coverage
Asia
Time Range
- Start Date: Not specified
- End Date: Not specified
Demographics
- Adults (where applicable)
- Mixed gender participants
- Industrial workplaces (60%)
- Residential environments (40%)
License
CC0 (Creative Commons Zero)
AI Training Rights
Licensee is granted a non-exclusive, worldwide, and perpetual right to:
- Use the Data Product to train, fine-tune, and evaluate machine learning models, including large language models.
- Incorporate Data Product content into models and commercialize resulting model outputs.
- Create derivative works (model weights, embeddings, feature representations, etc.) for any lawful purpose.
Restrictions
- The Data Product itself may not be sold, redistributed, or shared as a standalone dataset outside of licensed usage.
- Users must comply with all applicable privacy, data protection, and intellectual property laws.
Who Can Use It
- Data Scientists: Train stereo vision and perception models.
- Researchers: Conduct research in robotics, stereo vision, SLAM, and 3D perception.
- Robotics Engineers: Develop navigation, mapping, and autonomous robot systems.
- Businesses: Build AI-powered robotics and autonomous applications.
- Universities: Support education, benchmarking, and computer vision research.
Data Dictionary
| Column Name | Data Type | Description | Possible Values/Notes |
|---|---|---|---|
| Recording_ID | String | Unique identifier for each recording | Unique value |
| Left_Video | Video | Left camera video stream | MP4 |
| Right_Video | Video | Right camera video stream | MP4 |
| Calibration_File | File | Stereo camera calibration parameters | Intrinsic and extrinsic parameters |
| Environment | String | Recording environment | Industrial, Residential |
| Region | String | Geographic region | Asia |
| Camera_Device | String | Recording device | GoPro or equivalent |
| Resolution | String | Video resolution | 4K, 1080p |
| Frame_Rate | Integer | Recording frame rate | 30 or 60 FPS |
| Video_Codec | String | Video compression codec | H.264, H.265 |
| Audio | Boolean | Audio availability | Yes |
| Motion_Level | String | Motion intensity | Low to Medium |
| Depth_Data | Boolean | Native depth availability | No |
| Timestamp | DateTime | Recording timestamp | If available |
| Duration | Float | Recording duration | Seconds or minutes |
Additional Notes
- Recorded using synchronized stereo cameras based on a GoPro or equivalent system.
- Includes full stereo calibration parameters, allowing users to compute depth maps and disparity using standard stereo matching algorithms.
- Native depth maps are not included.
- Captured across industrial (60%) and residential (40%) environments in Asia.
- Videos are available in 4K and 1080p at 30/60 FPS using H.264/H.265 encoding with audio.
- Suitable for robotics, stereo vision, visual SLAM, autonomous navigation, 3D reconstruction, depth estimation, and embodied AI.
- Released under the CC0 license for both research and commercial AI applications.
TRUSTLoading...
£162,500
Download Dataset in Unknown Format
Recommended Datasets
Loading recommendations...