The aim of this project was to construct a system which allows the User to design and train a neural network to recognize handwritten digits, and then test the designed neural network model using uploaded images of digits handwritten by the User. The rationale for the application was to investigate the topic of image recognition and its applications by studying one of the simplest classification systems one can create using a neural network which is the recognition of handwritten digits.
Results from testing show the program successfully can train and test a personally designed neural network model. The program can successfully upload an image for analysis but appears to read the data incorrectly and as result it cannot make a confident prediction. Further work that could be carried out include creating a more sophisticated mount for the camera module to capture clearer images, as well as including the functionality of using the live video feed of a webcam in the project.
