Convolutional Neural Networks for Estimating the Ripening State of Fuji Apples Using Visible and Near-Infrared Spectroscopy

Abstract

The quality of fresh apple fruits is a major concern for consumers and manufacturers. Classification of these fruits according to their ripening stage is one of the most decisive factors in determining their quality. In this regard, the aim of this work is to develop a new method for non-destructive classification of the ripening state of Fuji apples using hyperspectral information in the visible and near-infrared (Vis/NIR) regions. Spectra of 172 apple samples in the range from 450 to 1000 nm were studied, which were selected from four different ripening stages. A convolutional neural network (CNN) model was proposed to perform the classification of the samples. The proposed method was compared with three alternative methods based on artificial neural networks (ANN), support vector machines (SVM), and k-nearest neighbors (KNN). The results revealed that the CNN method outperformed the alternative methods, achieving a correct classification rate (CCR) of 96.5%, compared with an average of 89.5%, 95.93%, and 91.68% for ANN, SVM, and KNN, respectively. These results will help in the development of a new device for fast and accurate estimation of the quality of apples.

Machine Learning for Infrared Spectral Classification of Polyvinyl butyrals with Identical Chemical Groups: An Example for Undergraduate Chemistry Classes

“…Additionally, CNN-based spectral classification models transform the original spectral data into higher-level expressions through a nonlinear model, finally learning complex features. The application of CNNs in infrared spectral analysis is a proven option among various deep learning algorithms used in spectral analysis. , …”

“…The application of CNNs in infrared spectral analysis is a proven option among various deep learning algorithms used in spectral analysis. 33,34 The MATLAB codes for the BPNNs and CNNs classification models are provided in the SI. Lecture Note 4 introduces BPNNs and CNNs.…”

Accurate felt-tip pen brands classification based on a convolutional neural network using data augmentation

“…To overcome this problem, Convolutional neural networks (CNN), one of the most popular deep learning, have been used to improve the accuracy of models. Deep learning algorithms reduce the reliance of raw data on pre-processing and feature ltering, can automatically extract features, and have shown excellent performance in the eld of spectral recognition [17]. However, deep learning requires extensive sample data for training.…”

Direct Identification and Quantitation of Protein Peptide Powders Based on Multi-Molecular Infrared Spectroscopy and Multivariate Data Fusion

“…It integrated one-dimensional IR spectroscopy, second derivative infrared (SD-IR) spectroscopy, and two-dimensional correlation IR spectroscopy (2DCOS-IR) with progressively higher resolution, and it employed to directly identify flour with diverse harmful substances, which is more suitable for the establishment of the model and has high accuracy . Chemometrics maximized the extraction of useful chemical information to achieve a better description and prediction of the natural sciences, which was widely used as a multivariate statistical analysis tool for food quality evaluation and identification, including adulteration identification, traceability, and grade evaluation − generally by building classification and regression models for qualitative and quantitative analyses, such as principal component analysis (PCA), support vector machine (SVM), k -nearest neighbor (KNN), back propagation neural network (BPNN), and partial least squares regression (PLSR). ,− …”