การประมวลผลภาพชุดข้อมูลตัวเลขด้วยโครงข่ายประสาทเทียมแบบคอนโวลูชัน (CNN) is licensed under a
Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 Thailand License.
This research aims to (1) evaluate the performance of the Convolutional Neural
Network (CNN) technique in analyzing and classifying handwritten digit images, and (2) compare
the performance of CNN with baseline techniques for digit image analysis. The Digit-MNIST
dataset, a standard dataset consisting of handwritten digit images from 0 to 9, was used for
experimentation. Classification models were developed using the CNN technique and baseline
methods. The results indicate that (1) the CNN technique achieved a classification accuracy of
98.64%, and (2) CNN outperformed the baseline techniques, where HOG + SVM achieved
95.75% accuracy and LBP + k-NN achieved 35.32% accuracy, respectively. These findings
demonstrate that the CNN technique is highly suitable and effective for processing and
analyzing digit images and can be effectively applied to other image recognition tasks.