Multi-modal versatility
Annotate conveyor video, drone imagery, and LiDAR together to cover sorting lines and sites.
Waste management
Teach models to sort waste and recover more.
A professional platform for real-time annotation of video, images, and sensor data. It produces datasets for AI that sorts waste, classifies recyclables, and optimizes recycling.
Applications
Automated waste sorting
Train sorting-line robotics to classify waste by material (plastic, paper, metal, glass) from annotated data.
Technical edge
AI-powered automation
Pre-label recurring materials on the belt, then interpolate them across fast-moving video.
Quality control
Automated checks and review catch misclassified materials before the dataset reaches training.
Collaborative workflow
Share high-volume labeling across annotators and reviewers, with role-based access.
Customization & integration
Define custom classes for materials and contaminants, and export to your sorting or ML stack via API.
Enterprise-grade security
Keep facility and operations data confidential, with on-premise deployment available.
Annotation types
Bounding box
Rectangular boxes that identify and classify conveyor items: bottles, cans, bags, and paper.
Oriented bounding box
Rotated rectangles that give robotic arms the position and orientation of complex-shaped objects.
Polygon
Flexible shapes that outline crumpled objects, organic waste, or contamination zones.
Points
Single points that mark container tags, small metal parts, or objects' centers of mass for robotic capture.
Lines & multilines
Trace sorting-area boundaries, conveyor lines, and collection-vehicle routes.
Skeletons
Connected keypoints that monitor personnel actions at sorting stations.
Cuboid
3D boxes that measure recyclable bale volume or container fill level.
3D point cloud
Label LiDAR data to analyze landfill volumes and map recycling facilities in 3D.
Semantic segmentation
Pixel-level classification of materials in the mass: plastic, paper, glass, metal, and hazardous waste.
Instance segmentation
Pixel-wise separation of each waste unit in a dense stream for counting and sorting.
Bitmap
Raster masks that isolate complex edges like shredded plastic or metal shavings.
Mesh
Polygonal meshes that build recycling-plant twins and model recycling processes.
Custom
A mix of annotation methods built around your specific waste-sorting or monitoring task.
Put your waste-management data to work
Book a demo and we'll scope your recycling dataset with you.
Custom solution? hello@keylabs.ai
