Can AI Hear the Sweetest Watermelon? Acoustic Detection Explained

Scene Description

China consumes about 70% of the world’s watermelons, with an average of 100 jin per person in 2018. A Zhejiang University PhD applied machine learning to judge watermelon maturity using acoustic features.

Keywords

watermelon, acoustic feature detection, machine learning

Why Acoustic Detection?

Traditional methods rely on farmers’ experience—observing skin color, texture, and knocking the fruit to listen to the sound. These methods are subjective, time‑consuming, and often inaccurate, especially for large‑scale production.

Machine‑Learning Approach

The researcher collected acoustic signals from knocked watermelons and extracted two moment‑based features (MI1 and MI2) that correlate with ripeness and internal hollowness. Four supervised algorithms were evaluated: Linear Discriminant Analysis (LDA), K‑Nearest Neighbors (KNN), Artificial Neural Network (ANN), and Least‑Squares Support Vector Machine (LS‑SVM).

Experimental Setup

Two experiments were conducted using "Kirin" watermelons harvested from a greenhouse:

Ripeness classification: 147 non‑hollow samples (75 for training, 72 for testing).

Hollow‑fruit discrimination: 190 samples (including hollow fruits) split into 97 training and 93 testing samples.

Results

LS‑SVM achieved the highest accuracy, with 73.6% correct classification on the test set, outperforming LDA, KNN, and ANN. The Fβ score (β=2) for LS‑SVM was 88.1% (training) and 74.7% (testing), indicating strong performance on imbalanced data.

Other algorithms showed lower accuracy: ANN (73.3% / 66.6%) and KNN suffered from bias toward majority classes, while LDA is suitable only for linear problems.

Advantages of Acoustic Detection

Compared with laser or nuclear‑magnetic‑resonance methods, acoustic detection is inexpensive, fast, and non‑destructive, making it suitable for large‑scale agricultural applications.

Conclusion

The study proves that AI‑driven acoustic analysis can reliably assess watermelon maturity and internal hollowness, providing a practical tool for growers to improve product quality and export potential.