ABCiFY: enhancing alphabet learning through image classification using Convolutional Neural Network (CNN)
Sophia Kaye G. Domingo, Clarissa V. Dominguez, Stephen E. Espeño, MARK ANTHONY S. MERCADO, Mary Grace D. Nulud, Ariel Antwaun Rolando C. Sison · IET conference proceedings. · 2025
Traditional teaching methods, such as flashcards are often not enough to make teaching and studying of Alphabets engaging for children. The study presents an innovative system website called 'ABCIFY', which is designed for alphabet image classification using Convolutional Neural Network (CNN), among children aged three to six. It enables children to capture common objects using their smartphones which reinforces children to engage in alphabet learning while at the same time familiarizing items in indoor environments (house and school)—for visual and kinesthetic learning. The web application features a Text-to-Speech function to support auditory learning, improving children’s pronunciation of letters. Additionally, it features an Interactive Quiz to evaluate and assist with children’s Alphabet recognition. The evaluation proceeded using ISO 25010, which determines and confirms the accuracy of the Image Classification in identifying objects and prompting responses, as well as the text-to-speech function for pronunciation and interactive quiz. The CNN model accuracy, on the other hand, demonstrated an 85% accuracy in image classification. To conclude, ‘ABCiFY’ reinforces a new approach of learning the alphabet in a more interactive and engaging way using Image Classification. Future enhancements of the study in terms of accuracy and functionality are essential for the improvement of the system.