Forest fire detection By Convolutional Neural Network:

Introduction :

• This Project Is To Implement Forest Fire Detection By Cnn .Used Basic Pretrained Model To Train Real Life Image and video For Classification and Object Detection.
• Libraries Used In This Project Keras,Matplotlib,numpy,Pandas,Computer Vision,Tensorflow,ImageIO.
• This Model Has Achieved 98 % Accuracy By Training And Testing the Model.
• This Is the First And Basic version Of this Project.

Project workflow :

• Download data - [https://github.com/prathyyyyy/Forest-Fire-Detection/blob/main/setup/setup_datasets.py]
• Data Analysis - [https://github.com/prathyyyyy/Forest-Fire-Detection/tree/main/Data%20And%20Performance%20Analysis]
• Neural Network - [https://github.com/prathyyyyy/Forest-Fire-Detection/blob/main/custom_model/cladoh.py]
• Pretrained Model -[https://github.com/prathyyyyy/Forest-Fire-Detection/blob/main/setup/naive_approach.py]
• Transfer learning and Data Augmentation - [https://github.com/prathyyyyy/Forest-Fire-Detection/blob/main/transfer_learning.py]
• Video Fire detection - [https://github.com/prathyyyyy/Forest-Fire-Detection/blob/main/video_annotation.py]
• Project Launcher - [https://github.com/prathyyyyy/Forest-Fire-Detection/commit/5452385558a01cc2cda4ec4d0f1febfbb0811f2a]

Major Libraries and framework :

1. Tensorflow and keras - This is an AI and Deep Learning library used in this project for creating neural network and pretrained model.
2. Numpy , pandas and matplotlib - This is used for data analysis , though for deep learning project it is not mandatory ,I have used just to give quick insight about data used in this project.
3. ImageIO , Computer Vision (opencv) - This library is used in this project to open , process the image and video .

Project tasks :

1. Download and setup dataset .
2. Perform data analysis for this project.
3. Create neural Network from scratch .
4. Setup pretrained model .
5. Data augmenting to extend dataset and enhance training.
6. By open-cv and imageIO open , proecess image and video.
7. Test and evaluate.
8. Launch the model.

Future Goals Regarding This Project

Code Refactoring & Simplification: The codebase will be refactored to improve modularity, readability, and documentation. The goal is to make the core concepts more accessible to beginners learning about deep learning and computer vision.

Integration of Advanced Models: We plan to leverage newer, more powerful pre-trained models like ConvNeXt-XLarge. This is expected to boost accuracy beyond 99% while significantly reducing training time and computational requirements.

Real-Time Object Detection: Transitioning from image classification to real-time object detection using a framework like YOLOv7/YOLOv8 or Detectron2. This will allow for precise localization of fire within a video stream, making the system suitable for live monitoring and alert systems.