Deep Learning based College Student Uniform Classification System using Transfer Learning

• Published in: 2024 IEEE 3rd World Conference on Applied Intelligence and Computing (AIC) pp. 861 - 866
• Main Authors: Suvarna, B, Balakrishna, Sivadi, Vinay, B, Siva, R, Charan, Y, Uday, P
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27-07-2024
• Subjects: Accuracy; Adaptation models; Computational modeling; D eep learning model; Deep learning; Feature extraction; Image Classification; Loss measurement; Reliability; Testing; Training; Transfer learning; Uniform Classification S ystem; VGG16
• DOI: 10.1109/AIC61668.2024.10731054

This study introduces a comprehensive classification system employing a deep learning network architecture to distinguish between student uniforms from different academic years. The system was designed to tell apart uniforms that appear somewhat similar. Images of uniforms from two distinct classes were collected from our college to create a dataset with uniforms from different years. The VGG16 model was utilized to extract features from these images. These features were compressed using a Global Average Pooling layer. The final features are then fed to deep learning model. This process achieved promising results with a notable 99% validation accuracy on the uniform dataset and 80.55% on shoe dataset. The system's high accuracy demonstrates the effectiveness of deep learning techniques in precisely classifying college uniforms, providing a reliable method for educational institutions to categorize various uniform types.