Benchmark Innovation Awards 2018 Site Protection (Software) Winner: Panasonic – WV-ASF950
Panasonic’s WV-ASF950 is facial recognition server software that utilises deep learning technologies. Featuring a core engine that boasts high facial recognition performance, the software can identify faces, including those that are difficult to recognise with conventional technologies. Trials have been shown where faces at an angle of up to 45 degrees to the left or right or 30 degrees up or down, and those partially hidden by sunglasses, have been accurately detected.
The software features an ‘iA’ (intelligent Auto) mode which automatically adjusts settings for the camera to capture optimal images which are best suited for facial recognition. When used with Panasonic’s
i-PRO Extreme series network cameras installed with a ‘Best Shot License Key’ which is provided bundled with the server software, only the most suitable images are transmitted to the server for facial recognition.
The combination of Panasonic edge devices and the facial recognition software maximises the performance of the core engine to achieve accurate and reliable recognition.
The roadmap for the software includes plans to add functionality which will enable recognition of partially covered faces; the manufacturer states this will be achieved by the end of 2018. Furthermore, cameras equipped with the iA function will ensure image analysis is performed at the camera instead of the server. This will result in reduced server and network loads, allowing system cost reductions.
The Deep Learning Face Recognition Software offers high levels of precision, and the iA function and Best Shot images maximise recognition engine performance.
It supports up to 30,000 faces (with an optional expansion pack) and integrated management with i-PRO monitoring enables the facial recognition system and video surveillance system to use the same GUI.
As demand for IoT and AI grows, the linking of various types of data is becoming increasingly beneficial, and is happening across many industrial sectors. Deep learning is being utilised to achieve this.
Panasonic has adopted deep learning technology in its facial recognition products. This enables the successful development of facial recognition technology to overcome the limitations of conventional systems.
An algorithm that combines deep learning with a similarity calculation method that suppresses errors enables recognition in situations that are difficult for conventional facial recognition technologies. The iA function detects scene conditions and optimises the settings accordingly to improve the detectability of the video images.