Research Article
Interactive Learning Media for Fruit Recognition in Early Childhood Using Backpropagation
@INPROCEEDINGS{10.4108/eai.18-11-2023.2342565, author={Zilvanhisna Emka Fitri and Siti Ingefatul Komariah and Lalitya Nindita Sahenda and Victor Phoa and Reski Yulina Widiastuti and Arizal Mujibtamala Nanda Imron}, title={ Interactive Learning Media for Fruit Recognition in Early Childhood Using Backpropagation}, proceedings={Proceedings of the 4th International Conference on Social Science, Humanity and Public Health, ICoSHIP 2023, 18-19 November 2023, Surabaya, East Java, Indonesia}, publisher={EAI}, proceedings_a={ICOSHIP}, year={2024}, month={1}, keywords={backpropagation computer vision fruit recognition interactive learning media}, doi={10.4108/eai.18-11-2023.2342565} }
- Zilvanhisna Emka Fitri
Siti Ingefatul Komariah
Lalitya Nindita Sahenda
Victor Phoa
Reski Yulina Widiastuti
Arizal Mujibtamala Nanda Imron
Year: 2024
Interactive Learning Media for Fruit Recognition in Early Childhood Using Backpropagation
ICOSHIP
EAI
DOI: 10.4108/eai.18-11-2023.2342565
Abstract
The challenges faced by early childhood education in rural areas include an inadequate number of teachers, inadequate facilities and infrastructure, and limited foreign language skills of students. To solve these problems, an interactive, easy, and interesting learning media was created to make students participate actively, help them recognise objects in foreign languages, and adjust the school need (especially schools with a limited number of teachers). Fruit was chosen as the subject of the study because students recognise various popular fruits but do not know their names in English. Computer vision with the backpropagation method was applied to classify and identify 11 types or 789 images of popular fruits. There are seven parameters learned such as red, green, blue, area, perimeter, shape, and diameter colour features. The optimal learning rate of 0.4 and maximum iterations of 500 resulted in a system accuracy rate of 100%.