kshitizb/pyrodetect repository overview

⁠🔥 PyroDetect - AI-Powered Wildfire Detection

PyroDetect is an advanced wildfire detection system using a custom Convolutional Neural Network (CNN). This AI-driven model enhances fire detection accuracy while minimizing false alarms, making it ideal for real-time deployment.

⁠🔹 Features

✅ Custom CNN architecture for precise wildfire detection
✅ Data augmentation techniques for robust performance
✅ Optimized training with Adam optimizer & batch normalization
✅ High accuracy and reduced false positives
✅ Fully containerized for seamless deployment

⁠📌 Installation & Usage

⁠Run the Container

docker run -p 8020:8020 kshitizb/pyrodetect

Then open: http://localhost:8020/⁠

⁠Pull the Image

docker pull kshitizb/pyrodetect

⁠📌 Input & Output

• Input: High-resolution aerial/ground wildfire images.
• Output: Predicted wildfire presence with confidence scores.

⁠📌 Model Performance

⁠Evaluation Metrics

• Confusion Matrix: TP: 216, TN: 129, FP: 30, FN: 35
• ROC Curve: AUC = 0.90
• Precision-Recall Curve: AUC = 0.99

⁠📌 Dataset

PyroDetect is trained on 2,700 high-resolution images collected from various sources:

• Government wildfire databases
• Public repositories (Flickr, Unsplash)
• Aerial and ground-based images

⁠🔹 Contributors

👨‍💻 Kshitiz Bhargava - Linkedin⁠ Github⁠

👨‍💻 Manvendra Singh - Linkedin⁠ Github⁠

👨‍💻 Abeer Mathur - Linkedin⁠ Github⁠