Fusarium Damaged Kernels Detection Using Transfer Learning on Deep Neural Network Architecture

TL;DR

This study applies transfer learning on a deep neural network to detect Fusarium-damaged kernels using a small dataset, achieving high accuracy with less specialized equipment compared to traditional hyperspectral imaging methods.

Contribution

It demonstrates that transfer learning on pre-trained DNNs can effectively detect Fusarium damage in wheat kernels with minimal training time and without specialized equipment.

Findings

Achieved 94.7% accuracy on a small dataset

Reduced training time to approximately 1 hour on a single GPU

Comparable accuracy to hyperspectral imaging methods

Abstract

The present work shows the application of transfer learning for a pre-trained deep neural network (DNN), using a small image dataset ($\approx$ 12,000) on a single workstation with enabled NVIDIA GPU card that takes up to 1 hour to complete the training task and archive an overall average accuracy of $94.7\%$. The DNN presents a $20\%$ score of misclassification for an external test dataset. The accuracy of the proposed methodology is equivalent to ones using HSI methodology $(81\%-91\%)$ used for the same task, but with the advantage of being independent on special equipment to classify wheat kernel for FHB symptoms.