Yandex Signal Q1 (the camera for the analysis of fatigue of the driver)

2019

Inclusion of algorithms SDK VisionLabs in a monitoring system of Yandex Signal Q1

On August 5, 2019 the VisionLabs company reported that analysis algorithms of persons of VisionLabs became a part of a monitoring system of fatigue and attention of drivers which Yandex intends to start . The taxi after pilot testing on several thousand cars. A system is capable to estimate the driver's condition then he in case of need receives the warning signal or access restriction to orders of service. The neuronet of Yandex learned to distinguish fatigue of drivers, and VisionLabs technologies helped to do it in the movement.

The device of monitoring of fatigue and attention is a small Yandex Signal Q1 infrared camera with specialized software. It is fixed on a windshield and directed towards the driver. The device was developed by specialists of Yandex. The taxi, it traces a status of the person driving on 68 key points on his face. At the same time it is required to distinguish persons as a fullface and in a profile, and in the movement, at a ducking — for this purpose were the SDK VisionLabs technology is used.

Algorithms SDK VisionLabs process each frame, find on is mute the person of the driver and issues coordinates of arrangement of eyes, a nose, a mouth and other key points. Data retrieveds are used for creation of the normalized frame of 250х250 in size where the person is strictly on the center. This frame can already be used for calculation of provision of the head in degrees on three axes: yaw, pitch and roll. To keep track of the status of eyes of the driver, algorithms analyze the image of eyes and for each eye make the decision: it is closed or it is open. Algorithms are able to define by means of IR Liveness technology whether there is in front of the camera a person or his facsimile. For the analysis the normalized frame is used, and at the exit the result of alive or notalive turns out.

After the analysis of key points of Yandex technology. Taxi determine fatigue degree by characteristic signs: to frequency and duration of blinking and also yawning and ducking. Processing and data analysis happens directly on the device therefore it can independently work without connection to the Internet. It allows to use the device in the conditions of unstable communication, for example in intercity buses or trucks. On the basis of the acquired information service make the decision on whether the driver can remain driving or to it the break for rest is necessary.

"It was technology difficult to learn to define the moment when degree of concentration of attention of the driver is close to decrease so that it is better for it to stop driving. For this purpose we needed to distinguish a set of points on the person and to analyze them using neuronets. We are glad that using colleagues from VisionLabs we managed to cope with this task", Roman Maresov, the head of Department of security of trips of Yandex noted. Taxi

"We are glad to provide a joint system for condition monitoring of drivers. The uniqueness of the project of Yandex is that all process happens in each camera, but not on separate servers", 'Alexander Hanin, the CEO of VisionLabs noted'

Announcement

On May 30, 2019 at a meeting on technologies in the field of artificial intelligence the managing director of "Yandex" Tigran Hudaverdyan showed to the president Vladimir Putin the device for the analysis of degree of fatigue of drivers.

Received  the name Yandex Signal Q1 the gadget represents the camera which, using possibilities of machine vision, controls the level of attention of the person driving.

"Yandex" showed to Putin the device for the analysis of fatigue of drivers

Electronic adaptation traces 68 points on the driver's person, performing monitoring of the direction of his look, turn of the head and frequency of blinking. If the driver distracts from the road, and the algorithm detects signs of fatigue or drowsiness, the camera gives to it a special signal: or to be focused on the road, or to take a break on trips.

The company tested Yandex Signal Q1 in cars of partners of service " Yandex.taxi ". Late "Yandex" is going to connect the AI device with the Taximeter application and to limit access to orders to those drivers of service who are inattentive or lost concentration. 

 Until the end of 2019 it is going to install Yandex Signal Q1 approximately on  5 thousand cars. Besides, with  the authorities of a number of regions negotiations on  installation of devices by  intercity buses and  in  city transport are conducted.

That " Yandex.taxi " is going to implement a similar monitoring system of a status of drivers,  it was reported in 2017. Then in the company claimed that investments into system development can make about $10 million.

Earlier in 2019 the company " Yandex.taxi " "Is lucky" and Citymobil announced data exchange system testing about cumulative operating time of drivers to exclude a possibility of processing of taxi drivers. Use of such technology is intended to reduce the risks connected with fatigue and to increase safety of transportations in Russia.^[1] 

Notes