×

You are using an outdated browser Internet Explorer. It does not support some functions of the site.

Recommend that you install one of the following browsers: Firefox, Opera or Chrome.

Contacts:

+7 961 270-60-01
ivdon3@bk.ru

  • Developing a machine learning model for estimating driver eye condition

    Currently, special attention is paid to artificial intelligence systems in transport. One of the actual directions is the development of self-driving car. However, there is also an intermediate approach. When the control is not fully automatic and automated. In such systems a human plays an important role, but the system, analyzing data from the environment, allows to form various kinds of recommendations. Moreover, such systems allow to work on the warning in some situations. In particular, some driver monitoring systems allow to detect the fact of smoking while driving, turning your eyes away from the road. In this article, special attention is paid to the system of analysis of the driver's eyes. First of all, the results of such analysis can be used to determine the fact of falling asleep. However, by analyzing, the frequency of blinks, it is possible to predict sleep. Nevertheless, computing power is needed to build such a system. In this work, the performance is achieved by using Haar cascades and the Viola-Jones method in eye detection. The eyes detected in the video are images of a much smaller size compared to the whole frame of the video sequence, due to which high performance is achieved in their processing by the convolutional network in the next step. Structurally, the system consists of two neural networks operating in parallel for the left and right eyes. The obtained values of completeness are about 90%.

    Keywords: computer vision, artificial intelligence, security system, face detector, deep learning, target skipping, proportion of correct recognition, driver monitoring, eye recognition, Viola-Jones method, Haar cascades, optimization

  • Studies of the average density and porosity of sulfur composite materials on quartz filler

    The parameters of the structure of the building material, including sulfur, depend on a large number of prescription and technological factors. This paper presents the experimental dependences of the average density and porosity of the sulfur composite material on the volume degree of filling, the specific surface of the filler, the concentration of the appret and the time of isothermal exposure. It is established that the dense sulfur composites obtained with silica filler with a specific surface area of 180 m2/kg, processed 30%-and 50%-s ' solution of rubber and subjected to isothermal aging for 2...3 hours

    Keywords: Sulfur, silica filler size, filler, volumetric filling degree, average density, porosity, dispersion

  • Specification of conditions of transport jams occurrence in networks with traffic light control

    In article process of vehicles queue formation before an adjustable intersection is considered. Classical models of delay and queue assume, that all vehicles arriving to intersection take places on a stop line. If in free conditions of movement the given assumption is comprehensible, in the conditions of the saturated movement the account of existential distribution of queue is required. For this purpose the description of queues dynamics is added by variables of queues front position and queues departure front position. Under concept the front of queue is understood position of last car in queue, and as departure front - position of the car beginning movement at travel of queue. Analytical dependences for calculation of values of the specified variables are resulted, both at constant arrival of vehicles, and at any cyclic profile of flow. Conditions of transport links blocking with the account of the entered variables are specified.

    Keywords: Delay, length of queue, transport links, blocking, traffic light control, front of queue, front of queues departure