Deep learning classification system for coconut maturity levels based on acoustic signals

TL;DR

The paper tackles non-destructive classification of coconut maturity using acoustic tapping signals. It augments an imbalanced dataset via audiomentation and procedural generation and trains two time-series deep learning models, RNN and LSTM, with MFCC-based 1D CNN feature extractors. Both models achieve 97.42% accuracy with no significant difference, and DL models outperform published ML baselines when evaluated on the augmented data. The approach demonstrates a viable, scalable path toward automated, non-destructive coconut maturity grading for export, with data quality and quantity driving margin improvements; future work includes collecting more samples and exploring additional features such as physicochemical properties.

Abstract

The advancement of computer image processing, pattern recognition, signal processing, and other technologies has gradually replaced the manual methods of classifying fruit with computer and mechanical methods. In the field of agriculture, the intelligent classification of post-harvested fruit has enabled the use of smart devices that creates a direct impact on farmers, especially on export products. For coconut classification, it remains to be traditional in process. This study presents a classification of the coconut dataset based on acoustic signals. To address the imbalanced dataset, a data augmentation technique was conducted through audiomentation and procedural audio generation methods. Audio signals under premature, mature, and overmature now have 4,050, 4,050, and 5,850 audio signals, respectively. To address the updation of the classification system and the classification accuracy performance, deep learning models were utilized for classifying the generated audio signals from data generation. Specifically, RNN and LSTM models were trained and tested, and their performances were compared with each other and the machine learning methods used by Caladcad et al. (2020). The two DL models showed impressive performance with both having an accuracy of 97.42% and neither of them outperformed the other since there are no significant differences in their classification performance.

Figures (4)

• Figure 1: A deep learning pipeline with training and testing phases
• Figure 2: Proposed model architecture
• Figure 3: Loss graph for RNN and LSTM models
• Figure 4: Confusion matrix for RNN model(left) and LSTM model (right)