Ente utilizes machine learning (ML) to enhance your photo experience while prioritizing privacy. Through on-device ML, your photos are analyzed locally on your device, ensuring data never leaves unencrypted. Despite challenges like limited compute and access to ML libraries, this approach guarantees privacy, lowers costs, and reduces latency. Unique features include Magic Search, utilizing Apple’s MobileCLIP model for semantic search, and face recognition using custom algorithms like linear incremental clustering. ML data is synced E2EE across platforms for a consistent experience. Ente expresses gratitude to the open-source ML community and hints at future enhancements with on-device ML while remaining open to exploring new techniques for user value.
https://ente.io/ml/