Abstract: Convolutional Neural Networks (CNNs) have revolutionized computer vision, achieving state-of-the-art results in facial recognition tasks. However, the performance of CNN models heavily depends on the quality and diversity of their training datasets. Existing datasets often lack sufficient representation of various ethnicities, leading to potential biases and limitations in real-world applications. This research explores the impact of using datasets specific to Indian ethnicity on the performance of CNN modules for facial recognition. By curating a diverse dataset that captures the unique facial features and variations within the Indian population, we aim to enhance the robustness and generalization capabilities of CNN models. Our proposed methodology involves collecting a comprehensive dataset, preprocessing the data, and training CNN architectures specifically tailored for Indian facial characteristics. Through extensive experiments and evaluations, we demonstrate the effectiveness of our approach in improving the accuracy and fairness of facial recognition systems for individuals of Indian ethnicity. This research underscores the importance of considering demographic diversity in dataset curation and model development, paving the way for more inclusive and unbiased computer vision applications.
Keywords: facial recognition, Indian ethnicity, training datasets.
|
DOI:
10.17148/IJARCCE.2024.13647
