<![CDATA[This study examines the application of the CNN method with the VGG19 architecture for digital image analysis in recognizing Indonesian sign language. The data used in this study is the BISINDO data set type, with 8,814 samples divided into 26 alphabetical categories. Implementing sign language recognition using the VGG19 architecture produces good accuracy results, reaching 93.24% with epoch 25 (without hyper-parameters tuning).These results confirm the model's extraordinary ability in image recognition and performing precise analysis. However, the results of this study can be improved again by performing Hyper parameters tuning on the architecture used, namely VGG19, by changing certain variables that affect increasing accuracy. Other aspects can be improved to achieve optimal performance, considering the excellent results. By integrating modern hyper-parameter tuning approaches and incorporating a variety of additional data, the model generalization is expected to be improved, leading to higher accuracy in many real-world settings]]>
Copyrights © 2025
