Abstract:
© International Research Publication House This article solves the problem of collecting initial data for constructing models for assessing the functional state of a person by pupillary response to changes in illumination. We analyzed the drawbacks of the traditional approach to collecting initial data using computer vision and time series smoothing methods. Attention is focused on the importance of the quality of the initial data for the creation of adequate high-precision mathematical models. We actualized the need for manual marking of the iris and pupil circles to improve the accuracy and quality of the initial data. We described the initial data collection stages in the proposed technology. We gave an example of the resulting pupillogram, which has a smooth shape and does not contain outliers, noise, anomalies and missing values. Based on the given technology, we developed a software and hardware complex, which is a collection of specially developed software that has two main modules and hardware implemented on the basis of a Raspberry Pi 4 Model B microcomputer, with peripheral equipment that implements the specified functionality. To evaluate the effectiveness of the given technology for collecting initial data, we used models of a single-layer perspetron and a collective of neural networks, together with the initial data on the functional state of intoxication of a person. The studies have shown that the number of errors of the 1st and 2nd genus in determining the assessment of the functional state of a person is lower, and the classification accuracy is higher when using the initial data generated by manual marking of circles, compared with the initial data collected by computer vision methods. Thus, the given technology for collecting initial data can be effectively used to build models for assessing the functional state of a person by pupillary response to changes in illumination.
