Histogram Layer for Texture Classification Jun 2019 - Aug 2019

Paper written and presented as part of the 61st Annual UF SSTP; Assisted ML texture analysis research in Professor Alina Zare’s Machine Learning and Sensing Lab by implementing deep network models in PyTorch and conducted various experiments with different parameters

We propose a hybrid model that incorporates a stackable, localized histogram layer on convolutional neural network (CNN) for texture analysis applications.

Instead of using standard histogram operation, we used RBF (Radial Basis Function) to perform a localized binning operation without binning constraints.


Media Features

Awards & Recognition

  • SSTP 2019Best Paper Award
© 2024 Tyler Kim. All Rights Reserved.