Crop-disease classification
Efficient CNN for real-time crop disease detection.
The preserved March 2026 manuscript reports a lightweight binary classifier trained on a balanced PlantVillage subset. It records 90.50% test accuracy, approximately 900,000 trainable parameters, a 3.6 MB model and CPU inference observations between 45 and 120 milliseconds.
These values are reported experimental results from the manuscript. Their public presentation should remain linked to the paper, dataset protocol and reproducible model evidence.
