Lightly

A computer vision suite that automates data curation, model pretraining, fine-tuning, and edge deployment. It solves the problem of building production-ready vision AI by helping ML teams select the most valuable data and pretrain models without labels. The suite targets ML engineers and organizations scaling computer vision, including Fortune 100 companies. It is delivered as a SaaS platform with open-source components, offering tools for data labeling, curation, and model training.

Key features

• Automated data curation at scale
• Pretrain vision models without labels
• Fine-tune models with knowledge distillation
• Smart data collection on edge devices
• Integrated labeling, curation, and QA
• Dataset management and versioning
• Self-supervised learning framework
• Standardized data labeling format