Panasonic Will Launch Facial Recognition Server Software

Washington Hispanic AP

Panasonic Corporation announced that it will launch a facial recognition server software that uses deep learning technology. The launch will be in July 2018 outside Japan and in August 2018 in Japan.

Thanks to a central engine that boasts the highest facial recognition performance in the world (* 1), this high-precision facial recognition software can identify faces that are difficult to recognize with conventional technologies, including faces at an angle of up to 45 degrees to the left or to the right, or up to 30 degrees up or down, and faces partially hidden by sunglasses. 

In addition, the new software features the “iA (Intelligent Auto) mode” (* 2), which automatically adjusts the camera’s settings to capture the most optimal images suitable for facial recognition. When used with the Panasonic i-PRO EXTREME series network cameras installed with the “Best Shot license key” that is included with the software, only the best shots, or “Best Shots”, will be sent to the server to facial recognition The combination of Panasonic’s basic devices and facial recognition software maximizes the performance of the software’s central engine for high-precision recognition. By the end of this year, the company plans to add a function to recognize faces partially covered by surgical masks, In addition, the use of this software with cameras equipped with the iA function allows to analyze images in the camera, instead of on the server, to send only the best images to the server. This will result in reduced loads for the server and the network, which results in a general reduction in system costs. In the event that 10 or more network cameras are connected to the system, costs can be reduced by approximately 40 to 50% compared to conventional systems that do not use the Best Shot function.

Panasonic will continue to improve its security-related products and provide a wide range of solutions to meet the increasingly diverse and changing needs of customers, such as facial recognition solutions for integrated management with a monitoring system. By providing these solutions, the company aims to become a “total integrator,” with the ability to contribute to customers’ frontline operations.
The facial recognition software that uses Panasonic’s deep learning technology has the following characteristics: 1. High precision:

– The most powerful facial recognition engine in the world according to NIST data (IJB-A facial challenge) – the iA function and Best Shot images maximize the performance of the facial recognition engine and provide high recognition accuracy
2. Reduced system cost: Send only the “Best Shot” images to reduce server and network load 
3. System expansion capacity:
– Register up to 30,000 faces (available as an option) – Integrated management with the i-PRO monitoring system: allows the integrated administration of a facial recognition system and a monitoring system with the same graphical user interface 

Notes:

* 1 In April 2017, the product reached the highest level of facial recognition performance in the world in a comparison test (Performance Report of IJB-A Face Verification Challenge / Performance Report of IJB-A Identification Challenge facial) of the National Institute of Standards and Technology (National Institute of Standards and Technology, NIST) of the United States, one of the most authoritative institutes in the world. * 2 A feature that uses Panasonic’s “LUMIX” mass-consumption digital camera technology to capture optimal images for facial recognition by automatically detecting moving objects, movement speed, faces and light intensity (either daytime or at night) found in video that are normally difficult to see due to the movement of the subject and the backlight, to allow the optimization of the configuration in real time.

[Fundamentals of development]
With the accelerated popularization of the internet of things and artificial intelligence, initiatives to connect various types of information to make life more convenient are being carried out in all industrial sectors. A new technology called “deep learning” is being used to achieve this. Panasonic has adopted this deep learning technology in its facial recognition products for the security industry. This has led to the successful development and commercialization of facial recognition technology that overcomes the difficulties of conventional technologies, such as recognizing faces when they are tilted, that have changed due to aging or are partially hidden by sunglasses.
[Product characteristics] 
High accuracy:
First level face recognition engine evaluated by NIST (IJB-A facial challenge) – The deep learning technology used in the new software was developed jointly with the National University of Singapore. The software improves facial recognition performance by up to 500% (* 3) compared to conventional systems. 

* 3 The rejection rate is reduced to 20% when the acceptance rate of the wrong person is set to 0.01 with the face image data set IJB-A.
– A unique algorithm that combines deep learning, an automatic learning method, with a similarity calculation method that suppresses errors, allows recognition in situations that are difficult with conventional face recognition technology, such as when the face is tilted (up to 45 degrees to the left or right, or 30 degrees up or down), partially hidden by sunglasses or a surgical mask (* 4), or that have changed by aging. 
* 4 Launch scheduled for the end of 2018
The iA and Best Shot functions maximize the performance of the facial recognition engine and provide high recognition accuracy – The iA function allows cameras to detect the scene and automatically optimize settings to improve the detection capability of video images. The camera automatically detects moving objects, movement speed, faces and light intensity (day / night / headlights) found in the video and which are normally difficult to see due to subject movement and backlight, and optimizes real-time settings and capture the most optimal videos of the subject.

– The Best Shot license key included with this software can be installed on the Panasonic i-PRO EXTREME series network cameras (sold separately) to automatically select the appropriate images for facial recognition of the multiple images of Faces captured when a person passes in front of the camera and sends only those selected images to the server. This allows high quality images suitable for facial recognition to be sent without loading the server.

2. Reduction of the cost of the system:
Send only Best Shot images to reduce server and network load
– With conventional facial recognition systems, all captured images are sent to the server and the server performs face detection and facial recognition, which concentrates the load of data processing on the server. The systems also tend to be large scale due to the large bandwidth required to send all the images and the large amount of hard disk space required to save those images.
– By using the Panasonic system that combines this product with cameras that include the iA function allows the camera to detect the faces and only send the Best Shot images containing faces to the server, which drastically reduces the network load and eliminates the need for large bandwidth. Performing face recognition on the server using Best Shot images reduces server load and hard disk capacity requirements, as well as allowing up to 20 * 5 network cameras to connect to a single server.

– With this Panasonic system, it is no longer necessary to send high resolution images, which consume a large amount of data, to a server, unlike conventional systems. This provides high precision facial recognition while drastically reducing network traffic, transmission costs and network construction costs.

– When using the Panasonic system with 10 or more network cameras connected, costs can be reduced by 40 to 50% compared to conventional systems that do not use the Best Shot function.
* 5 The number of cameras that can be connected depends on the number of people passing through the camera and the retention period for the facial images.

3. System expansion capacity:
Register up to 30,000 faces (the option is sold separately)
Simple lot registration of a maximum of 10,000 faces is possible with the standard face recognition server software (WV-ASF950). The optional facial registration expansion kit (WV-ASFE951W) allows the registration of up to 30,000 faces in large-scale installations.
Integrated management with the i-PRO monitoring system
Facial detection and recognition and tracking with recorded video can be performed in the same graphical user interface through integrated management with the client software WV-ASM300 or WV-ASE231W for Panasonic i-PRO monitoring systems. This eliminates the need for the dedicated operation and administration required for conventional facial recognition systems, and allows the work to be centralized. 
Specifications 
– Facial recognition server software
Product number

WV-ASF950 WV-ASF950W
General description
Number of faces: Up to 10,000 faces can be registered. This number can be expanded to a maximum of 30,000 faces by using the facial registration expansion kit. 
Number of cameras: Up to four cameras can be connected.
This number can be expanded to a maximum of 20 cameras using the existing camera expansion kit WV-ASFE901 (x 1) and WV-ASFE904 (x 4).
-Facial record expansion kit
Product number
WV-ASFE951W
General description
Expands the number of faces that can be registered from the standard number of 10,000 to 20,000 (with a license) or 30,000 (with two licenses).