OMRON Electronic Components Europe has released a new OKAO Vision face recognition package, allowing highly accurate deep-learning face recognition. Developers can deploy OKAO Vision flexibly on their own preference of embedded hardware platform.
The new deep learning libraries of OKAO Vision Face Recognition V9.0 address applications that need a high degree of certainty under different conditions, including poor lighting and when the face is at different angles relative to the detector. An important possible application is monitoring the participation at face-face and online meetings, aiding contact tracing and verifying actual attendance. Automotive applications incorporate driver recognition to handle features such as seat adjustment.
The new face-recognition libraries deliver excellent evaluation results with different skin tones and face sizes, with very low error rate down to image size as small as 40 pixels. Benchmark testing with Intel and Arm processors has shown that the solution maintains particularly fast recognition times despite the enhanced accuracy, assuring that users in access control applications, for example, will be hardly conscious of the necessity to wait for validation of their identity.
The complete package contains modular libraries that provide various sensing capabilities comprising age and gender estimation, expression estimation, and photographic image beautification incorporating red-eye reduction, eye enlargement, facial shaping, and blemish removal. Users can consolidate various modules’ functionalities to add value to their applications.
Gabriele Fulco, European product marketing manager Sensors at OMRON Electronic Components Europe added: “In addition to boosting recognition accuracy with deep learning, we are making OKAO Vision available as a set of software libraries ready to integrate with Linux, Windows, and iOS operating systems. This lets users leverage the extensive functionalities of OMRON’s machine-vision package in their own embedded systems running on custom hardware. Off-the-shelf libraries are already available for various platforms.”