This project focuses on the development of a machine learning model for the multi-class classification of date fruits. Utilizing TensorFlow and Keras, we've constructed a Sequential neural network tailored for this task.
The dataset used in this project is the "Date Fruit Datasets" available on Kaggle. It consists of images of various types of date fruits, suitable for multi-class classification tasks.
- Accessing the Dataset: To access and download the dataset, visit Date Fruit Datasets on Kaggle.
The dataset undergoes several preprocessing steps to ensure optimal model performance:
- Feature Scaling: Min-Max scaling is applied to normalize the feature values.
- Label Encoding: Target variables are encoded to facilitate model training.
- Standardization: The StandardScaler from Scikit-learn is used to standardize features, achieving zero mean and unit variance.
The neural network model comprises:
- Model Architecture: A Sequential model with five dense layers, including dropout regularization to prevent overfitting.
- Activation Function: ReLU is used in hidden layers, with a Softmax activation in the output layer for multi-class classification.
- Optimizer: Stochastic Gradient Descent (SGD) with Batch Normalization, selected for its efficiency and performance, showing a test accuracy of 91.85%.
To replicate or further develop this project, ensure you have TensorFlow, Keras, and Scikit-learn installed in your Python environment. The model can be trained and evaluated using the provided scripts, with customization options available for model architecture and training parameters.
Contributions to improve the model or extend its functionality are welcome. Please submit pull requests for review.
See the LICENSE file for details.
- TensorFlow and Keras Teams for providing the deep learning framework.
- Scikit-learn Contributors for preprocessing tools.