Abstract:
The object of this study is a procedure for measuring physical quantities
under laboratory conditions at educational institutions. The issue related
to this case is the lack of any comprehensive method and technical solution
suitable for the experimental study
of physics in both offline and online
learning formats. To solve this problem, an approach has been proposed,
based on computer vision technology
and training special neural models to
recognize objects in video frames that
perform mechanical movement.
The idea of the proposed approach
is based on the hypothesis that by measuring the position of an object in video
frames with sufficient accuracy, it is
possible to determine the functional
type of the law of its motion. Further,
knowing the function of the law of
motion, it is possible to calculate any
physical quantities describing the process under consideration. The idea is
implemented in the form of a technical
solution, which is a set of prototypes of
automated laboratory devices.
The choice of the method for determining the law of motion was carried
out using the analysis of the recognition error, measurement error, speed
and resistance to external conditions
of the Hough algorithmic method and
the YOLOv8n neural network model.
It is shown that the neural network
method YOLOv8n has very high accuracy but low performance. The Hough
method shows high performance but
lower accuracy and resistance to external conditions. It was found that the
accuracy of the YOLOv8n method is 4
times higher, but the execution speed is
10 times lower than that of the Hough
method. However, in the case of artificial lighting and fixing the distance
from the camera to objects, the Hough
method provides 99.9% accuracy in recognizing an object in video frames.
The obtained prototypes of devices can be used for further research to
determine their impact on the quality
of physics education
