Classification for handwritten characters on MNIST dataset using CNN and LOOCV evaluation

This repository contains the final semester project for the Machine Vision course, which contains handwritten character classification using the EMNIST dataset. The project utilizes Convolutional Neural Networks (CNN) enhanced with EfficientNet through transfer learning to speed up the training process. Evaluation is performed using Leave-One-Out Cross-Validation (LOOCV), featuring metrics such as confusion matrix, accuracy, precision, and F1-score.

Result

Training Result

Here is a preview of the image that is displayed based on the prediction and true label:

Picture 1: CNN Accuracy

Evaluasi CNN

Model evaluation results using the Convolutional neural network (CNN) method:

Picture 2: CNN Confusion matrix

Evaluasi LOOCV

Model evaluation results using the Leave-One-Out Cross Validation (LOOCV) method:

Picture 3: LOOCV Confusion matrix