PlantVillage Plant Disease Detection (EfficientNet)
This model is a Keras implementation of EfficientNet trained to classify 38 types of plant diseases from the PlantVillage dataset.
Model Description
- Architecture: EfficientNet (Keras .keras format)
- Task: Image Classification (38 classes)
- Dataset: PlantVillage (color images)
Evaluation Results
The model achieves an overall accuracy of 41.68% on the provided test set.
Metrics
- Accuracy: 41.68%
- Total Test Images: 511
Intended Use
This model is intended for educational and research purposes in the field of agricultural plant pathology. It can be used to identify potential diseases from leaf images.
Limitations
- Performance may vary significantly across different plant species and disease types.
- The model was trained on the PlantVillage dataset and may not generalize well to images taken in different environmental conditions or on different leaf backgrounds.
How to Use
import tensorflow as tf
from tensorflow.keras.models import load_model
import numpy as np
# Load model
model = load_model("plant_disease_efficientnet.keras")
# Preprocess image (EfficientNet expectation)
# ... see inference.py for details