Excel Sheet Classification with Keras

This project trains a neural network model to classify Excel sheets into four categories:

• Summary (0)
• Transaction (1)
• Others (2)
• Cheque (3)

Installation

1. Clone the repository or download the script.
2. Install dependencies using:

   pip install -r requirements.txt

Usage

1. Place your .xlsx files inside the train-excel/ directory.
2. Run the script:

   python train.py

3. The trained model will be saved as trained_model.h5.
4. The scaler used for normalization will be saved as scaler.pkl.

Model Training

• The script extracts features from each Excel sheet.
• It normalizes data and trains a neural network with 64-32-4 dense layers.
• Early stopping is enabled to prevent overfitting.
• The final accuracy of the model is printed after evaluation.

Output

• trained_model.h5 → Saved Keras model
• scaler.pkl → StandardScaler object for data normalization

Evaluation

After training, the model is evaluated using test data, and the accuracy is displayed.

Inference

1. Load the trained model model = keras.models.load_model("trained_model.h5") in inference.
2. Load the new .xlsx file inside test-excel/ directory.
3. Run the script:

   python infer.py

Result

Final result excel will be saved in predicted_sheets.xlsx .