# AutoAnotationModel

This repository contains the custom model checkpoint `Model.pth` and a powerful annotation pipeline developed by **Rajkumar G** for real-time object auto-annotation and YOLO dataset generation.

## 🔧 What This Is

`AutoAnotationModel` enables:
- ⚙️ Multi-class object detection using template-based region matching
- 🧠 Intelligent segmentation refinement using custom logic
- 🖼️ Live visual preview of annotations through a Flask web application
- 📤 YOLO-format `.txt` export for training detection models
- 📦 Exportable annotation packages including image + labels

The model and its web interface are used in applications like:
- Bin/Cover detection
- Mechanical part annotation
- Industrial dataset creation
- Real-time inspection and labeling

---

## 🧠 Model Information

- **Model Name:** `Model.pth`
- **Hosted on:** [Hugging Face - Rajkumar57/AutoAnotationModel](https://huggingface.co/Rajkumar57/AutoAnotationModel)
- **Created by:** Rajkumar G
- **Frameworks Used:** Python, PyTorch, OpenCV, Flask
- **Use Case:** Real-time annotation & YOLO label generation

---

## 🚀 How to Use the Model

```python
from huggingface_hub import hf_hub_download
from segment_anything import sam_model_registry, SamPredictor
import torch

# Download model from Hugging Face Hub
model_path = hf_hub_download(repo_id="Rajkumar57/AutoAnotationModel", filename="Model.pth")

# Load model
model = sam_model_registry["vit_h"](checkpoint=model_path)
model = model.to("cuda" if torch.cuda.is_available() else "cpu")
predictor = SamPredictor(model)
```

---

## 🌟 Key Features of My Annotation System

- ✅ Web interface for user-defined class names and class IDs
- 📁 Upload template images for each class
- 🔍 Automatic region detection and segmentation
- 📤 YOLO-format `.txt` annotation generation
- 📸 Annotated image previews
- 📦 One-click export as `.zip`

---

## 📁 Files Included

| File        | Description                                  |
|-------------|----------------------------------------------|
| `Model.pth` | Custom-trained model for annotation tasks    |
| `README.md` | Documentation for usage                      |

---

## 👨‍💻 Author

**Rajkumar G**   
📧 g.p.rajkumar5@gmail.com  
🔗 [LinkedIn](https://linkedin.com/in/rajkumar) | [GitHub](https://github.com/rajkumar)

---

## 📜 License

This model and pipeline were fully created by **Rajkumar G** and are released strictly for **academic and research purposes only**.

For commercial use, integration support, or contributions, feel free to reach out directly.

---

*Proudly developed and maintained by Rajkumar G. If you use this project, kindly give credit or cite this repository.*