Rank One Computing has released version 1.19 of the ROC SDK, its flagship facial recognition solution, featuring accuracy improvements across all use cases and important new features.
- Improved Facial Recognition Accuracy – Version 1.19 delivers across-the-board accuracy improvements, particularly on unconstrained imagery and young and old people.
- Improved Facial Clustering – Massive improvements have been delivered to our facial clustering algorithm. Coupled with the speed of the SDK, users can now accurately and rapidly cluster identities across large datasets.
- GPU-accelerated Template Generation – In addition to Rank One’s industry-leading enrollment speed on CPUs, users can now optionally perform enrollment on a GPU and achieve up to a 3x speed up.
- Homomorphic Encrypted Matching – Encrypt templates and perform matching in the encrypted space, without the template ever being decrypted, via Rank One’s new patent-pending method. In turn, higher levels of security and privacy in biometric transactions can be enabled between trusted and unknown parties.
- Facial Hair Estimation – Rank One’s facial analytics algorithms now detect the presence and thickness of beard and mustache hair.
- Video Tracking API Enhancements – Another round of improvements has been delivered to API tools for tracking and identifying faces in video.
A detailed description of the improvements delivered in this release can be found in our blog post.