The aim of the study is to assess accuracy spatial object localization in mono and stereo modes visual-to-auditory sensory substitution by means developed system tested on persons with normal or corrected-to-normal vision.Materials Methods.A prototype a device based video camera two lenses was prepared.Software convert signal from into an audio developed.To system, experimental 30 blindfolded sighted participants conducted.15 were mode, 15 -in mode.All trained use system.During experiment,...

Sensory substitution represents an opportunity to experience visual-like perception for the visually impaired people. However, effects of sensory perception, and, in particular, depth are less known. The study aimed evaluate influence one monocular cues-linear perspective, on accuracy object localization using vOICe visual-auditory technology. 32 sighted volunteers, who were blindfolded during experiment, participated study. sample was divided into 2 groups, which used a linear perspective...

Accurate information on air quality serves as the foundation for making regulatory and legal decisions aimed at reducing pollution. This study investigates vertical distribution of dust particle concentration, their elemental composition, size in atmospheric surface layer Yekaterinburg. Over eight days April 2021, 64 samples were collected filters heights ranging from 0.5 m to 10 a single site using mobile post. The mass concentration dust, characterized by heterogeneous data with...

Retrospective information about the state of urbanized environment can be obtained by analyzing images located in Internet space. Artificial intelligence-based methods allow recognizing objects a given class images, while models based on neural networks demonstrate high recognition accuracy. In this paper, network model its own dataset for an is proposed using example construction works creation geoinformation system with coordinate time reference. The pixellib library was used image...

Sensory substitution devices (SSDs) provide visually impaired users with a "visual" perception of the environment by transmitting information about field view head-mounted camera. To increase effectiveness SSDs, visual data should be pre-processed to convey spatial object in simple and understandable way. In this work, authors supposed that use stereo approach would give user more accuracy localization task. The experiment involved 30 sighted participants, who were blindfolded during...

The paper proposes an original approach for predicting the values of spatial series. This can be used, in particular, to recover missing data. counter-prediction method was tested on a model artificial neural network (ANN), which is sequentially trained preceding predicted segment series left and right. final prediction weighted average results these two sets. We have work using example dust content snow cover. 256 samples were taken with step 0.2 m along line area dumps existing open pit...

Retrospective information about the state of urbanized environment can be obtained by analyzing images located in Internet. Methods based on artificial intelligence allow recognizing objects a given class images, moreover models neural networks demonstrate high recognition accuracy. In this paper, network model is proposed its own dataset for an example classes construction works, asphalt damage and gravel to get retrospective data panoramas with different time capture. The pixellib library...

The article is devoted to the problem of choosing a representative selection subset for an artificial neural network in tasks interpolation distribution metals topsoil. Environmental data, often used build models, are datasets at irregular points. traditional division input data into training and test subsets occurs randomly, which transfers number problems. For subset, question individual collective representativeness points asked, sending them request on content element soil given area....

Many objects of the real world, both living and man-made, may be described in terms graph theory represented as data. One tasks data analysis is to classify topologies. This work explored possibilities machine learning methods, particular neural networks (GNNs), graphs with different The aim was study robustness classifiers based on convolutional depending samples they were trained on. authors also compared solutions GNNs non-graph classifiers, which fed several characteristics. included...

The study of the dynamics greenhouse gases in Arctic regions planet is becoming increasingly important. Such studies are especially relevant due to climate change observed this region. paper propose a hybrid model that combines wavelet transformation original data and an artificial neural network with long chain short-term memory (LSTM) elements predict changes surface methane concentration latitudes. time series via discrete transform was decomposed into four components — one approximating...

The selection of a method for dividing the raw data into training and test subsets in models based on artificial neural networks (ANN) is an insufficiently studied problem continuous space-time field interpolation. In particular, selecting best subset modeling spatial distribution elements topsoil not trivial task, since sampling points are equivalent. They contain different amount “information” point each specific model, therefore, when modeling, it advisable to use most containing...

The article proposes the use of permutation method for assessment predictive ability models based on artificial neural networks. To test this method, three networks were implemented: a multilayer perceptron, radial basis function network, and generalized regression network. For modeling, data spatial distribution copper iron in topsoil (depth 0.05 m) territory subarctic city Noyabrsk, Yamalo-Nenets Autonomous Okrug, Russia, used. A total 237 soil samples collected. modelling, concentration...

A number of studies showed that the form presentation stimuli affects perception these and their processing by brain. In categorization tasks, different categories in visual verbal forms was compared. However, mechanisms various require additional research. The paper considers application psychophysical model signal detection theory (SDT) to problem categorizing textual pictographic stimuli. SDT is used measure sensitivity, biases discriminability performed tasks. An experimental study...