Services for the data you already have

Capabilities

• Object detection and segmentation
• 3D bounding boxes and 6D pose
• Hand and finger keypoints
• Trajectories and gripper state
• Contact and force
• Action, task and timing labels
• Language instructions and scene graphs
• Model-assisted labelling for image, video, 3D and OCR
• Delivery in LeRobot, MCAP or RLDS
• Three-tier quality checks aimed at 95% or better accuracy

Capabilities

• Redacts faces, screens, documents, licence plates, badges and tattoos in footage
• Removes personal data from documents
• Lets you choose to mask, blur, hash or scramble, with an audit record for each job
• Stores no footage on their servers
• Adds quality checks and documentation to go with redacted datasets

Capabilities

• Quality scores per operator, task and episode
• Feedback while collection is still running, so teams improve as they go
• Works with teleoperation, handheld‑gripper and first‑person collection
• Takes MCAP, ROS bags, LeRobot or raw video with control logs
• Tells you whether a problem came from the operator or from the capture

Capabilities

• Robotics annotationfirst‑person footage, hand detection and keypoints, how people handle objects, scene understanding, LiDAR and 3D point clouds
• Image annotationboxes, classification, segmentation, keypoints
• Video annotationframe by frame, tracking several objects, activities and events
• Human quality review on every delivery