About Lightly
Lightly is a computer vision suite designed for ML teams and organizations scaling vision systems. It covers the full workflow—from data curation to model pretraining, fine-tuning, and edge deployment—helping teams ship production-ready CV faster.
Key Features
In practice, Lightly stitches data handling, model training, and deployment into an integrated workflow. It helps teams move from raw data to deployed CV with less labeling and manual overhead:
Data Curation & Labeling
Integrated platform to curate, label, QA, and manage CV datasets; open-source at its core and scalable for ML engineers and organizations.
Unlabeled Pretraining
Pretrain vision models for industrial applications without labeled data, enabling faster transfer learning.
Edge Data Collection & Selection
Real-time, high-value data collection on edge devices reduces data transfer and storage needs.
Dataset & Model Management
End-to-end tooling for dataset versioning, labeling QA, and managing models for production.
Summary
Best for ML engineering teams, data scientists, and MLOps teams deploying computer vision in production.
Pricing
View pricingCommunity
Free / Monthly
- Active Learning • Uncertainty
- Similarity
- Balancing Analytics • Visualizations
- Reports
- Statistics
- Data Storage • AWS S3, Google Cloud, Azure, and more
- Data Type • Images (up to 1024px)
- Automation • API
- Tooling integration
- Data export
- input dataset limit: 25k
Custom
Tailored / Yearly
- Team plan, plus: • On-prem/priv. cloud install
- Production Data • Edge/Cloud selection
- Model/Drift monitoring
- Data Type • Video (4K UHD and more)
- Images (4K UHD and more) • Sequences
- Object level (Crops)
- Special datatypes
- Unlimited • Data/Teams
- Security/Training • SLA/Security
- 2FA/SSO
- Training
