Convolutional Neural Networks and TensorFlow-Based Enhance Correction for Color Visual Deficiency (CVD) Impaired

Color Vision Deficiency (CVD) presents significant challenges for individuals, particularly in distinguishing certain colors under normal lighting conditions. The primary objective of this research is to enhance the autonomy of people living with CVD by developing an innovative application. Leveraging Convolutional Neural Networks (CNNs), our solution aims to assist users in accurately identifying and correcting colors for CVD. The methodology encompasses collecting and preprocessing image data, followed by designing and implementing the system utilizing TensorFlow. Data was collected from open-source datasets and web scraping, then preprocessed by resizing to 224x224 pixels, and augmenting with random flips and rotations. The CNN model was designed with multiple convolutional, pooling, and fully connected layers, trained with the Adam optimizer, batch size of 16, and 35 epochs. The system underwent rigorous training and evaluation, yielding a promising accuracy of 90.49% on training data and 88.13% on test data. Our work stands out by providing a practical, user-friendly tool specifically designed to address the daily needs of those with CVD. This research not only enhances our understanding but also contributes to developing more effective strategies and tools to support and improve the quality of life for those affected by color vision impairments.