Recognition and identification of the clothes in the photo or video using neural networks
A. Ivanov, Georgii I. Borzunov, Konstantin Kogos · 2018
We introduce our solution of the problem of clothing recognition and identification. The cause of taking convolutional neural network as a mathematic model for the further research was the result of an experimental survey of methods of keypoints detection and computing their descriptors. The convolutional neural network model was developed and trained on the Apparel Classification with Style clothing image dataset. The training image dataset was expanded with the help of augmentation methods. The computational experiment aimed at determination of neural network efficiency and stability proved its productivity. Moreover, we developed clothing color recognition, which let us perform clothing identification by 2 parameters: type and color.