Klasifikasi Tingkat Kematangan Buah Tomat Menggunakan K-Nearst Neighbor (K-NN)

This research evaluates the performance of the K-Nearest Neighbor (KNN) algorithm in classifying tomato ripeness levels using 6D HSV color features [Hμ,Sμ,Vμ,Hσ,Sσ,Vσ]. The tomato dataset (green, breaker, turning, red) was processed through a production pipeline: Google Drive mounting, intelligent batch loader (batch_size=30, max 100/class), RGB histogram spectral analysis, adaptive preprocessing (R>100 filter + 3×3 morphology), and HSV feature extraction with
memory-efficient garbage collection. Evaluation on 16 test
samples achieved 93.75% accuracy via confusion matrix: 100% recall/precision for ripe class (8/8 correct), 87.5% for unripe class (7/8 correct, 1 false positive). Clear spectral progression was measured: Green (Hμ=100°) →Red (Hμ=15°, R=230), with declining G/R ratio as maturity metric.
KNN proved reliable for perfect ripe tomato detection without missed cases, with only 1 minor error in green transition samples. The system is ready for farmer edge
device deployment to optimize screenhouse harvest timing, demonstrating HSV color features + KNN effectiveness in tropical smart agriculture.
Keywords: KNN, HSV 6D, tomato maturity, confusion matrix 93.75%, spectral progression, batch processing, edge deployment