| Mechatronic System for Handwritten Digit Recognition Using CNN |
| Paper ID : 1098-IUGRC5 (R1) |
| Authors: |
|
Mohamed Ashraf Eldahshan1, Omar Fathy Eltokhy1, Abdelrahman Salah Zaghloul *1, Ehab Mohamed Said2 1Printing Engineering Department, Military Technical College, Cairo, Egypt 2Printing Engineering Department, MTC |
| Abstract: |
| Nowadays, all aspects of life call for automating dull and repeating jobs. Computer vision & deep learning are widely used in the automation process. This paper demonstrates the use of deep learning to facilitate collecting and monitoring student grades for the members of the examination control room. The main goal of the experimental setup is automating the process of handwritten digit recognition. Several methods of deep learning were used in the field of handwritten digit recognition like Support Vector Machine (SVM), Convolutional Neural Network (CNN), K-nearest neighbor (KNN) & Multilayer Perceptron (MLP). From the previous research, CNN was found to achieve higher accuracy; as a result, it was adopted in this research. The famous handwritten digit dataset called the Modified National Institute of Standards and Technology database (MNIST) is utilized in this research. The accuracy of the CNN model was verified using the testing dataset. After that, several processes are developed like interfacing python with Arduino, interfacing python with excel, QR code detection, and digital image pre-processing. Finally, the recognition process was successfully validated by testing the experimental setup on real-life pictures. |
| Keywords: |
| Mechatronics, Deep learning, handwritten digit recognition, Convolutional Neural Network (CNN). |
| Status : Paper Accepted (Oral Presentation) |
