Open AccessReducing dimensions of the histogram of oriented gradients (HOG) feature vector
Author(s) -
Stanislav Shidlovskiy,
A. S. Bondarchuk,
S Poslavsky,
M. V. Shikhman
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1611/1/012072
Subject(s) - support vector machine , pattern recognition (psychology) , histogram , artificial intelligence , histogram of oriented gradients , feature vector , pixel , feature (linguistics) , computer science , classifier (uml) , computer vision , convolution (computer science) , image (mathematics) , artificial neural network , linguistics , philosophy
The article discusses the basic principles of a HOG-descriptor. A method is proposed for reducing dimensions of the feature vector obtained using the HOG-descriptor by applying the convolution operation and converting the resulting values of gradients and their directions. After using new feature vectors for training, the Support Vector Machine (SVM) classifier showed a 37-fold increase in performance when processing an image with a resolution of 282×159 pixels.