ultra_cnn
A set of Python scripts for using Keras and TensorFlow to develop and assess "face recognition" methods based on convolutional neural networks for identifying individual tree logs, as described in the publication E. Holmström, A. Raatevaara, J. Pohjankukka, H. Korpunen, and J. Uusitalo, "Tree log identification using convolutional neural networks", Smart Agricultural Technology (2023) (https://doi.org/10.1016/j.atech.2023.100201). Code for the classical LBP approach presented in the paper is also available here.
The dataset used in the publication can be found on Zenodo:
https://doi.org/10.5281/zenodo.7281916
Uses the following packages: tensorflow, os, matplotlib, shutil, subprocess, sys, glob, time, numpy, warnings, math, random, sklearn, skimage
Contact: eero.holmstrom@luke.fi