About

This Application streamlit detect flames and smoke into 📷Picture, 🎥Video and 💻Webcam using Transfer Learning based on Yolov8x. Each image/frame (encoded into binary) and its detection (bounded boxes, labels, confidence...) are saved into a mongodb

Install 🏗

1. Clone the repository where you want
2. Install requirements.txt pip install -r requirements.txt
3. Create a MongoDB call Fire_Detection with 3 collections : Image, Video and Webcam
4. Launch a terminal and navigate to the folder cloned (normally call 'fire_detection') 🫠
   4.1 Navigate to the folder app
   4.2 Launch the streamlit application with the command streamlit run Home.py
5. Enjoy the experience
6. Consult your MongoDB to get the track of your detection

Screenshot 🖼

Next Features 🔮

• Upgrade the saving in DB (currently image-frame multiple detection -> one entry per detection | Future : image-frame with multiple detection -> one entry (all labels / bounding boxes)
• Page to watch the saving data
• Video detection from an url
• Saving the detection on your computer (Button download), not only in DB
• Possibility to change weight / model
• Add a confidence cursor
• Add a loader during processing

Cleanup for your own usage

You can safely remove the following files/folders from repo root:

• .github/
• LICENSE
• README.md
• CHANGELOG.md