🧾 AI_Image_Detection

Receipt Data Detection with YOLOv11

This project detects key information such as seller_name, seller_id, total_items, and amount from input receipt images using YOLOv11 and deep learning.

---

📁 Project Structure

My Model V11/
├── main.py              # 🧠 Main program for image detection
├── best.pt              # 🏆 Trained YOLOv11 model
├── yolo11n.pt           # ⚙️ YOLOv11 base model used for training
├── output_images/       # 🖼️ Folder to save predicted images
├── model/               # 📊 Training outputs
│   ├── confusion_matrix.png
│   ├── P_curve.png
│   ├── results.csv
│   └── weights/
│       ├── best.pt
│       └── last.pt
├── DataSets/            # 📂 Dataset structure for training
│   ├── data.yaml
│   ├── train/
│   ├── valid/
│   └── test/
└── labelme/             # 🏷️ Labeling files created with LabelMe

---

🛠️ Tools & Technologies

• 🔍 Object Detection: YOLOv11
• 🏷️ Labeling Tool: LabelMe
• 🖼️ Input: Receipt images
• 📤 Output: Bounding boxes with classified receipt fields
• 💻 Training: Google Colab (T4 GPU)
• ☁️ Storage: Google Drive

---

🚀 How to Use

1. Label your data using LabelMe and export as YOLO format.
2. Organize your dataset into train, valid, and test folders under DataSets/.
3. Train your model in Google Colab using T4 GPU and YOLOv11:
   • Use yolo11n.pt as the base weights.
4. Run detection by executing main.py.
5. Check results:
   • Predictions are saved in output_images/
   • Training performance is visualized in model/ with confusion matrix, precision-recall curve, and results log.

---

🧠 Sample Output

Predicted receipts will display bounding boxes labeled with:

• Seller Name 🏪
• Seller ID 🆔
• Total Items 📦
• Amount 💵

---

📌 Notes

• Make sure your Google Drive is mounted properly in Colab.
• Adjust data.yaml to reflect your class names and path settings.
• Keep main.py updated to use best.pt for accurate inference.

---

📫 Contact

If you'd like to contribute or have questions, feel free to reach out or open an issue!

---