TAQWA: Teaching Adolescents Quality Wadhu/Ablution Contactlessly using Deep Learning

This research presents a unique and innovative approach to teaching young children the proper steps of ablution (wazoo/wudu) by utilizing a non-invasive sensing system integrated with deep learning algorithms. However, most existing ablution detection systems rely on cameras, which raise privacy concerns, face challenges with lighting conditions, and require complex training with long video sequences. We conducted experiments with a group of youngsters to evaluate the system’s effectiveness, demonstrating its potential in fostering a deeper appreciation and comprehension of religious practices among young learners. This innovative privacy-preserving ablution system employs state-of-the-art UWB-radar technology with advanced Deep Learning (DL) techniques to effectively address the challenges mentioned above. The core focus of this system is to categorize the four fundamental ablution steps: Wash Face 3x, Wash Hand 3x, Wash Head 1x, and Wash Feet 3x. By transforming the collected data into spectrograms and harnessing the sophisticated DL models VGG16 and VGG19, the proposed system accurately detects these ablution steps, achieving an impressive maximum accuracy of 97.92% across all categories with the utilization of VGG16.