Macroalgae Auto-Label + Coverage Web Tools

🧩 Core Features

Features	Description
📤 Upload new underwater images	User can upload new image(s) via the web interface
⚙️ Auto-label with YOLO + SAM	YOLO detects → SAM segments each detection
🖼️ Display results interactively	Show segmentation overlays + class-wise masks
📊 Calculate % coverage	Pixel-based area per class (for each image)
💾 Download annotations & coverage	Export JSON or image+mask zip archive

🛠️ Tech Stack

• Frontend + Backend: Streamlit (modern, fast, interactive UI)
• Model Backend: Your fine-tuned YOLO model + pretrained SAM model
• Image Ops: OpenCV & NumPy
• Optional Storage: Save results locally or integrate with Firebase/GCP later.

📁 Directory Structure

project/
├── custom-yolo/           # folder containing pre-trained YOLO model but .gitignored
├── images/                # raw images to label
├── output/
│   ├── masks/             # generated masks
│   ├── overlays/          # visualization of predictions
│   └── annotations.json   # structured result
├── sam-model              # contains the Segment Anything Model
└── scr/                   # contains notebook
|   ├── app.py
├── readMe.md
├── requirements.txt

📂 To Run It Locally:

• Save this file as app.py
• Place yolo_model.pt and sam_b.pt in the same directory
• Run the requirement file

pip install -r requirements.txt

• Install dependencies:

pip install streamlit opencv-python pillow ultralytics

• Run the app:

streamlit run project/scr/app.py