ResNet50 Fine-tuned for Wildfire Detection
This model is a ResNet50 fine-tuned on the Wildfire Prediction Dataset for image classification to detect wildfires.
Model Details
The model is a ResNet50 architecture with a custom classification head. It was trained in two phases: first with the backbone frozen, and then with full fine-tuning.
Training
The model was trained on the provided dataset with the following characteristics:
- Train samples: 30250
- Validation samples: 6300
- Batch size: 16
- Epochs (Phase 1 - Head): 5
- Epochs (Phase 2 - Full Fine-tuning): 3
Performance
(Add details about the model's performance metrics, e.g., accuracy, loss, etc. from your training output)
Usage
(Provide code examples on how to load and use the model for inference)