Real-world datasets,
at AI scale.
We execute targeted field data capture campaigns across India to power spatial computing, bimanual robotics manipulation, and computer vision models. Double-opt-in consent logs, 7-14 day mobilisation, and full GDPR compliance.
Ongoing capture campaigns & egocentric datasets
A real-time snapshot of active first-person (egocentric) POV and robotics training data collection drives across India.
Active
Industrial Egocentric POV Capture
Continuous first-person video capture of manual labor workflows in textile and garment manufacturing units. Utilizing high-resolution head-mounted cameras to record fine-grained hand-object manipulation, tool usage, and sewing machine operations in natural factory settings.
Active
Multi-Camera Gaze & Facial Tracking
High-fidelity dataset capturing driver and operator attentiveness using head-mounted eye-tracking rigs. Simultaneously records dual-eye infrared gaze vectors and external scene views, training attention-mapping and safety-critical ML models for autonomous transport.
Active
Humanoid Robotics Manipulation Dataset
Rigorous bimanual manipulation data capturing hand poses, finger joint tracking, and object grip configurations. Subjects wear sensor-equipped smart gloves and wrist-tracking modules during multi-step sorting, folding, and assembly tasks, feeding training loops for next-generation humanoid hands.
Active
Egocentric Daily Task & Appliance Interaction
First-person camera datasets of common domestic tasks and smart appliance operations. Features egocentric video recording of cooking, cleaning, and meal preparation under varying indoor lighting conditions. Designed to train home-assistant robotics.
Targeted acquisition in complex real-world environments
No synthetic data, no web-scraping — every dataset is captured by trained field operatives from real consented participants under rigorous QA audits.
👁️ Egocentric / First-Person POV
Chest rigs, head mounts, and eyewear capturing commutes, manual labor, and domestic tasks—the exact visual coordinates your robotics models require to learn.
🚶 Human Behaviour Video
Subjects walking, cooking, shopping, driving, and interacting across urban, semi-urban, and rural Indian environments. Visual complexity at scale.
🤖 Robotics & Manipulation Data
Fine-grained hand-object interactions, bimanual sorting, and tool usage across textile units, agricultural fields, and warehousing operations.
🏷️ Annotation & Labelling
Skeletal keypoint landmarks, bounding boxes, polygon segmentation, temporal boundary tags, and frame metadata mapping.
😊 Activity & Gesture Vocabulary
Multi-camera capturing of gestures, micro-expressions, and physical trajectories across diverse age, gender, and regional demographics.
🎙️ Speech & Dialect Audio
Acoustic data recorded in natural settings (home, street, work) across 10 major Indian languages to close the ASR accent gap.