🪙 Ancient_Coin_Authentication_Classifier

In this model/Classifier, Convolutional Neural Network (CNN) model based on VGG16 was used to classify the real or fake Ancient Bengal Coins. Convolutional Neural Network (CNN) model based on VGG16 to classify coins as real or fake

🚀 Features

✔️ Pretrained VGG16 Model – Transfer learning for better accuracy
✔️ Binary Classification – Predicts real 🏅 or fake ⚠️ coins
✔️ Python & TensorFlow – Implemented using deep learning frameworks
✔️ Dataset Processing – Preprocessing, augmentation, and normalization

🛠️ Technologies Used

🖥️ Python – Core programming language
📦 TensorFlow / Keras – Deep learning framework
📊 Matplotlib – For visualization
📁 OpenCV – Image processing

📂 Dataset

The dataset consists of images of real and fake coins. Preprocessing includes resizing, normalization, and data augmentation to improve generalization.
Used dataset was collected form a research team.

🛠️ Training and Evaluation

✔️ Data augmentation applied to enhance generalization.
✔️ Model trained on Google Colab.
✔️ Performance evaluated using accuracy, loss curves, and confusion matrix.

📌 How to Use

1️⃣ Clone the repository:

git clone https://github.com/ArunRoy404/Ancient_Coin_Authentication_Classifier.git

2️⃣ Install dependencies:

pip install os
pip install numpy
pip install matplotlib
pip install tensorflow
pip install notebook

3️⃣ Install flask:

pip install flask

3️⃣ To train the model:

code "Ancient Coin Authentication Classifier.ipynb"

4️⃣ To run the model using flask:

python app.py

📊 Model Performance

✔️ Accuracy: 98.17% 🏆
✔️ Loss: 6.91% 📉
✔️ Confusion matrix 📊

🚀 Future Improvements

✔️ Expand dataset for better generalization.
✔️ Implement real-time coin classification using a webcam.smartphone application.
✔️ Improve model performance using other CNN architectures like ResNet or EfficientNet.