AUTHOR=Wang Yi , Song Shuran 

TITLE=Detection of sweet corn seed viability based on hyperspectral imaging combined with firefly algorithm optimized deep learning

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 15 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2024.1361309

DOI=10.3389/fpls.2024.1361309

ISSN=1664-462X

ABSTRACT=Hyperspectral imaging technology combined with deep learning networks is widely used for identifying seed vitality grades. However, the settings of these neural network hyperparameters largely depend on the trial and error of the operator. To address this problem, in this paper, a firefly algorithm (FA) optimized CNN-LSTM (FA-CNN-LSTM) is proposed to detect sweet corn seeds viability grade through hyperspectral images. First, the hyperspectral images of 496 seeds, including four viability-grade seeds, are extracted and preprocessed. Then, the objective function for optimizing the hyperparameters of the CNN-LSTM model using the FA algorithm was defined. Finally, several classic neural network models and machine learning algorithms are compared with the FA-CNN-LSTM model. The experimental results indicate that the FA-CNN-LSTM model proposed in this study outperforms other models with a classification accuracy of 97.23%, 2.64% higher than the lowest-performing CNN, and 1.29% higher than CNN-LSTM. Furthermore, the FA-CNN-LSTM has a more significant advantage than traditional machine learning algorithms. Therefore, optimizing neural network hyperparameters using optimization algorithms is feasible and effective.