STEM approach prototype to improve classification skills in tea leaves using convolutional neural networks

The STEM approach integrates Science, Technology, Engineering, and Mathematics in technology-based problem solving. This study aimed to develop a valid, practical, and effective STEM-based prototype to improve students’ classification skills in identifying tea leaf diseases using Convolutional Neural Networks (CNN). The study employed a Research and Development (R&D) method using the ADDIE model. The developed products included Student Worksheets (LKM) and Learning Outcome Tests (THB). The results showed that the prototype was valid, practical, and effective, with validation, practicality, and effectiveness percentages of 94.18%, 90.33%, and 94.28%, respectively. Students’ classification skills improved significantly, as indicated by the increase in average scores from 49.40 to 80.77 and the paired sample t-test result of 0.000 < 0.05. Qualitative analysis using phase portraits and N-Vivo also demonstrated improvements in classification skill indicators. Therefore, the STEM-based prototype effectively improved students’ classification skills in CNN-based tea leaf disease identification.