Aleksey V. Zinkevich
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Aleksey V. Zinkevich has a total of 1 co-authors. Here are the top co-authors by number of shared publications:
- S. V. Sai
Pacific National University
2 shared publications
Latest Publications
- Open Access
- 1 February 2025
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 25 (1) , 68
Method of semantic segmentation of airborne laser scanning data of water protection zones
This article presents an evaluation of the efficiency of a neural network method for the semantic segmentation of three-dimensional point clouds obtained using the Geoscan 401 Lidar UAV. The proposed implementation of the neural network is based on the PointNet++ deep learning model which directly processes point clouds. A technique has been developed for acquiring and preparing a dataset with four classes: land, vegetation, vehicles, and construction objects. To increase the accuracy of the evaluation, a technique based on augmentation and redistribution of the datasets has been proposed. The neural network model consists of hierarchically constructed blocks that perform sampling, grouping, and feature extraction. Adjusting the number of blocks and setting the search radius for local features affects both the accuracy of segmentation and computational costs. The efficiency of the method for semantic segmentation of three-dimensional point clouds obtained using the Geoscan 401 Lidar UAV has been evaluated. The augmentation and redistribution technique improved the average Intersection over Union (IoU) value by at least 35 %. For the obtained data, the optimal radius in the grouping layer was determined, ensuring a balance between detail and sensitivity. It was found that an increase in the number of points in the dataset does not lead to a significant improvement in accuracy; however, the diversity of the datasets used enhances the method efficiency. The developed dataset increases the effectiveness of the applied approach, including when training other models. The results of this study indicate the potential for using the proposed methods and algorithms in constructing digital models of the Amur River and its main tributaries.
- 1 February 2025
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 25 (1) , 68
Open AccessMethod of semantic segmentation of airborne laser scanning data of water protection zones
This article presents an evaluation of the efficiency of a neural network method for the semantic segmentation of three-dimensional point clouds obtained using the Geoscan 401 Lidar UAV. The proposed implementation of the neural network is based on the PointNet++ deep learning model which directly processes point clouds. A technique has been developed for acquiring and preparing a dataset with four classes: land, vegetation, vehicles, and construction objects. To increase the accuracy of the evaluation, a technique based on augmentation and redistribution of the datasets has been proposed. The neural network model consists of hierarchically constructed blocks that perform sampling, grouping, and feature extraction. Adjusting the number of blocks and setting the search radius for local features affects both the accuracy of segmentation and computational costs. The efficiency of the method for semantic segmentation of three-dimensional point clouds obtained using the Geoscan 401 Lidar UAV has been evaluated. The augmentation and redistribution technique improved the average Intersection over Union (IoU) value by at least 35 %. For the obtained data, the optimal radius in the grouping layer was determined, ensuring a balance between detail and sensitivity. It was found that an increase in the number of points in the dataset does not lead to a significant improvement in accuracy; however, the diversity of the datasets used enhances the method efficiency. The developed dataset increases the effectiveness of the applied approach, including when training other models. The results of this study indicate the potential for using the proposed methods and algorithms in constructing digital models of the Amur River and its main tributaries.
- Open Access
- 1 February 2023
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 23 (1) , 79
Computational methods to increase the speed of FPGA-based discrete wavelet transforms
The article considers the computational methods and features of the construction of a complex functional block for the implementation of the discrete wavelet transform (DWT) Dobeshie 9/7 in digital image signal processing systems based on FPGA. We proposed a mathematical model and algorithms for the implementation of parallel and series-convector methods of signal processing to calculate the coefficients of a discrete bi-orthogonal Dobeshie wavelet 9/7 taking into account the architecture of used FPGA. The model is based on wavelet transform factorization methods using lifting schemes. In contrast to conventional lifting schemes, the proposed method and algorithms can increase the speed of FPGA calculations with simplified hardware implementation. CAD Quartus II and ModelSim are used as a development environment. The behavioral model is written in Verilog HDL. Altera Cyclone® IV 4CE115 was used as FPGA. On the basis of the obtained behavioral model the testing module was developed and the simulation of digital circuit in the ModelSim environment was carried out. The formula for estimating the number of clock cycles of the forward and reverse DWT has been proposed; on its basis the estimate of the number of parallel computations depending on the number of input elements and the characteristics of the FPGA was obtained. As a result of experiments the dependences of the number of cycles for DWT computation depending on the size of the side of a square image with different variants of the number of parallel processing blocks were obtained. It is shown that parallel work of several independent modules gives a possibility to conduct concurrent processing of several input columns (rows) from input 2D array, and unification of the multiplier-summing module allows to increase efficiency of calculations and to reduce volume of occupied hardware resources. Conveyor based DWT structure is characterized by less hardware costs in terms of implementation of the calculator unit and memory allocation. As a result of testing the digital circuit, it was found that the developed block structure can significantly increase the DWT speed as well as reduce the cost of the system on a chip. The proposed realization of the block of two-dimensional forward and reverse wavelet transform for the Dobeshi 9/7 filter bank forms a complete module and can be used as a ready-made complex functional block for further development of high quality image transmission systems in real time.
- 1 February 2023
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 23 (1) , 79
Open AccessComputational methods to increase the speed of FPGA-based discrete wavelet transforms
The article considers the computational methods and features of the construction of a complex functional block for the implementation of the discrete wavelet transform (DWT) Dobeshie 9/7 in digital image signal processing systems based on FPGA. We proposed a mathematical model and algorithms for the implementation of parallel and series-convector methods of signal processing to calculate the coefficients of a discrete bi-orthogonal Dobeshie wavelet 9/7 taking into account the architecture of used FPGA. The model is based on wavelet transform factorization methods using lifting schemes. In contrast to conventional lifting schemes, the proposed method and algorithms can increase the speed of FPGA calculations with simplified hardware implementation. CAD Quartus II and ModelSim are used as a development environment. The behavioral model is written in Verilog HDL. Altera Cyclone® IV 4CE115 was used as FPGA. On the basis of the obtained behavioral model the testing module was developed and the simulation of digital circuit in the ModelSim environment was carried out. The formula for estimating the number of clock cycles of the forward and reverse DWT has been proposed; on its basis the estimate of the number of parallel computations depending on the number of input elements and the characteristics of the FPGA was obtained. As a result of experiments the dependences of the number of cycles for DWT computation depending on the size of the side of a square image with different variants of the number of parallel processing blocks were obtained. It is shown that parallel work of several independent modules gives a possibility to conduct concurrent processing of several input columns (rows) from input 2D array, and unification of the multiplier-summing module allows to increase efficiency of calculations and to reduce volume of occupied hardware resources. Conveyor based DWT structure is characterized by less hardware costs in terms of implementation of the calculator unit and memory allocation. As a result of testing the digital circuit, it was found that the developed block structure can significantly increase the DWT speed as well as reduce the cost of the system on a chip. The proposed realization of the block of two-dimensional forward and reverse wavelet transform for the Dobeshi 9/7 filter bank forms a complete module and can be used as a ready-made complex functional block for further development of high quality image transmission systems in real time.
- 17 May 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 670-674
ESP8266 Microcontroller Application in Wireless Synchronization Tasks
The operation of UDP protocol in the Wi-Fi network and Espressif ESP-NOW protocol is studied. In the first case, an external Wi-Fi access point is used to communicate between two ESP8266 microcontrollers, and in the second case, both microcontrollers operate as paired devices. The standard libraries of the Arduino IDE programming environment are used in the source codes. The local time of sending and receiving is saved during the packets exchange, on the basis of which the processing time and delay of the packet, as well as, the system clock discrepancy of both ESP8266 microcontrollers are calculated. To obtain experimental data, 5000 packets are exchanged. The evaluation is carried out by applying various operating frequencies of the microcontroller (80 MHz and 160 MHz), as well as the operating library version of API for ESP8266 application (Non-OS SD 2.2.1 and Non-OS SDK pre-3). The obtained results demonstrate the best values for short packets when ESP-NOW protocol is used, however, the deviation can be obtained in almost 25% of cases even with absence of transmission/reception error.
- 17 May 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 670-674
ESP8266 Microcontroller Application in Wireless Synchronization Tasks
The operation of UDP protocol in the Wi-Fi network and Espressif ESP-NOW protocol is studied. In the first case, an external Wi-Fi access point is used to communicate between two ESP8266 microcontrollers, and in the second case, both microcontrollers operate as paired devices. The standard libraries of the Arduino IDE programming environment are used in the source codes. The local time of sending and receiving is saved during the packets exchange, on the basis of which the processing time and delay of the packet, as well as, the system clock discrepancy of both ESP8266 microcontrollers are calculated. To obtain experimental data, 5000 packets are exchanged. The evaluation is carried out by applying various operating frequencies of the microcontroller (80 MHz and 160 MHz), as well as the operating library version of API for ESP8266 application (Non-OS SD 2.2.1 and Non-OS SDK pre-3). The obtained results demonstrate the best values for short packets when ESP-NOW protocol is used, however, the deviation can be obtained in almost 25% of cases even with absence of transmission/reception error.