Datasets:
metadata
license: cc-by-nc-4.0
task_categories:
- video-classification
- robotics
tags:
- egocentric
- sim-to-real
- agriculture
- physical-ai
- computer-vision
Banas AI: areca_plant_pot
Asset Title: Multimodal Real-World Egocentric Video & Manipulation Dataset for Embodied AI
Banas AI is a specialized data provider supplying real-world, first-person (egocentric) physical datasets for artificial intelligence development. We focus on capturing human manipulation, spatial navigation, and complex physical tasks from a first-person perspective.
Target Use Cases
- Humanoid Robot Training & Sim-to-Real Transfer
- Large Multimodal Model (LMM) & Video Foundation Model Training
- Spatial Computing, AR/VR Gesture & Environmental Understanding
- Autonomous Navigation and Obstacle Avoidance
Modalities & Technical Specs
- Video Format: High-Definition MP4/MKV
- Sensor Streams: 6-DoF IMU telemetry, LiDAR depth maps, eye-tracking vectors.
- Annotation Types: Unlabeled raw video, timestamped action labels.
- Hardware Sources: Meta Ray-Ban, GoPro, Custom head-mounted rigs.
Legal & Compliance
All data is ethically collected with explicit, signed biometric consent. Incidental PII is automatically blurred. Small sample datasets are distributed under CC BY-NC for evaluation purposes.
*For enterprise licensing, contact contact@banas-ai.com or visit *https://www.banas-ai.com/