Deep Learning Tracker to track the trajectory of the selected object. Undergraduate work.
This file (main.tex) is launched using XeLaTeX compiler.
Keywords: object tracking, quantization of weights, deep learning network, number of quantization levels, neurocomputer network.
The object of the study is a network, which refers to the type of deep training of a neurocomputer network.
Purpose of the work: research, development and modification of the existing algorithm for tracking a moving object on a set of images in real time. It was achieved by modifying the DLT deep trust network using a weighting quantization algorithm.
As a result, the number of quantization levels necessary for the quantization algorithm and acceptable tracking is obtained; An analysis of the effect of quantization on tracking an object is given.
I advise you to read the article on the DLT tracker and the article on the weighting coefficient quantization algorithm before starting work.
Also, you can familiarize yourself with the interesting work of comparing trackers and with sequences designed specifically for such programs.
- *Tlepbergenova Darya * - MMF NSU 2020
See also the code of the program analyzed in this work.