Gessen Pavel Alekseevich (Postgraduate student
Peter the Great St.Petersburg Polytechnic University
)
Saradgishvili Sergei Erikovich (Candidate of technical sciences, associate professor
Peter the Great St.Petersburg Polytechnic University
)
| |
This work is devoted to the problem of the quality of object identification and tracking in smart home ecosystems with the possibility of video processing. Due to the limitations in this area, the object of interest is approaches that remain computationally efficient with weak provision of computing resources. The authors conducted a study of both proven approaches and the latest methods in the field of tracking and identification of objects. Previously, the authors proposed a system that complements existing high-performance implementations to improve the accuracy characteristics of tracking and detection by adding scale and partial rotational invariance, a classical approach to extracting key features due to a histogram of directed segments, the support vector method as a classifier and further training it in the process. In the current work, an analysis of possible approaches to improving the previously presented method is carried out, existing solutions to the problem of collecting and determining key features of an object are studied, the implementation of a system with a neural network approach to their identification is provided, as well as the results of the work of original algorithms and using both proposed solutions, which show an improvement in the accuracy indicators of both tasks of interest. These improvements can be used in any smart home systems, although the possible scope of application may be much wider, to improve the quality of video surveillance, which may have a better impact on the quality of life of consumers of such products: from becoming a smart home even “smarter” in terms of predictive ability to improving security by improving surveillance systems.
Keywords:training; support vector machine; neural networks; clustering, tracking.
|
|
| |
|
Read the full article …
|
Citation link:
Gessen P. A., Saradgishvili S. E. SYSTEM FOR IMPROVING THE QUALITY OF OPERATION OF OPTICAL-ELECTRONIC COMPLEXES OF A SMART HOME // Современная наука: актуальные проблемы теории и практики. Серия: Естественные и Технические Науки. -2025. -№04. -С. 55-60 DOI 10.37882/2223-2966.2025.04.07 |
|
|
LEGAL
INFORMATION:
