The problem PrismieDigitPredictor solves

Our model is a small step towards building a fully functioned handwriting recognition model.
In Handwriting Recognition (HWR) the device interprets the user's handwritten characters or words into a format that the computer understands (e.g. Unicode text). The input device typically comprises a stylus and a touch-sensitive screen.

Our model can correctly identify Handwritten digits from 0 to 9.
It uses pattern matching to convert handwritten letters into corresponding computer text or commands in real time.
A larger modified version of our model can be used to read number plates at traffic signals, toll plazas. It can also be used to read handwritten digits in postal code and recognising digits in bank cheques. It can be applied/used for surveillance, healthcare and various other sectors.

The applications of digit recognition include in postal mail sorting, bank check processing, form data entry, etc. The main problem lies within the ability on developing an efficient algorithm that can recognize hand written digits, which is submitted by users by the way of a scanner, tablet, and other digital devices.
One of the advantage of Handwriting recognition is Better data storage. Handwriting recognition paves the path for optimal data storage. Many files, contracts, and personal records include handwritten information, such as original signatures or notes, that can be converted into electronic text with handwritten text recognition technologies.

Challenges we ran into

-> Inadequate knowledge about pygame, keras and few other methods/packages. We learned about the required packages and used them in our model.
-> Wrong predictions while using Tkinter GUI inspite of the model being 98% accurate. Hence, we used pygame GUI to take the inputs and predict the result. We were able to get correct predictions by the model using this package.
-> Error of mismatch of size of output provided by the model and target data even after no syntactical errors. Later we realised it was due to running the part of the code creating CNN model more than once, that leads to doubling of layers. We rectified our mistake and hence came up with an error free model.