Lightly

About Lightly

Lightly provides tools that select the most informative images and video frames for labeling, reducing annotation volume, and supports self-supervised pretraining of vision models. It offers both a hosted product and open-source libraries. It is used by teams building perception and inspection models.

Common use cases

• Labeling cost reduction
• Dataset curation
• Model pretraining

Products and services

Key features

• Data selection
• Duplicate removal
• Self-supervised pretraining
• Embedding search
• Open-source libraries

Integrations

• AWS
• Google Cloud
• Labelbox
• CVAT

Lightly alternatives

• Voxel51Visual AI data tooling built around the open-source FiftyOne project
• EncordData infrastructure for annotating, curating and aligning multimodal AI data
• RoboflowComputer vision platform for labeling, training and deploying vision models
• V7AI tooling for data labeling and automation of document and image workflows

FAQs