This repository contains a Python script designed to augment image datasets, particularly in preparation for training deep learning models. The script uses TensorFlow to apply a variety of random transformations to images, increasing the diversity of the dataset and improving the robustness of machine learning models.
- Random Transformations: Includes horizontal flipping, random rotation, brightness, contrast, saturation, and hue adjustments.
- Noise Addition: Introduces Gaussian noise to the images for further variability.
- Random Translation: Slight shifts in the image position to simulate different perspectives.
- High-Quality Output: Augmented images are saved in the same format and quality as the original.
- Batch Augmentation: Processes entire directories of images, creating multiple augmented versions of each image.
- Data Preparation for GANs: Ideal for projects like WGAN-GP, where diverse training data is crucial for generating high-quality images.
- Image Classification: Enhances datasets for training more robust and generalized models in image recognition tasks.
- Research and Development: Useful for researchers and developers experimenting with data augmentation techniques.
- Clone the repository.
- Update the
dataset_pathvariable in the script to point to your image dataset directory. - Run the script to generate augmented images.
- TensorFlow
- Python 3.x
This project is licensed under the MIT License - see the LICENSE file for details.