This model uses Neural Networks and Random Forests to determine if the input photo contains my handwriting or not.
Important : Since I cannot upload the photos here, hence I have made available a trained model in the jupyter notebook. Thus, please run the notebook from the 5th last line. All the above lines in the notebook are for training the model if the photos can be uploaded
All the data used has been taken by my phone and hence needed some preprocessing and data augmentation. Neural networks don't seem to work well as they give some 60% accuracy at max on the train data while Random Forests give 100% on train and 80% on test data.
Model_for_prediction.ipynb : This is main jupyter notebook file written in Python 3
model.txt : This text file contains one of my saved models for Random Forests
my.jpg, my1.jpg, my2.jpg: Test pictures that are my handwriting (You can't find them anywhere else ;)
not_my.jpg, not_my1.jpg : Test pictures that are not my handwriting (You can try some for your own)
photos.txt : Contains the address for all the images that are my handwriting (But sadly Github doesn't allow me to upload more than 100 pictures)
not_photos.txt : Contains the address for all the images that are not my handwriting (But sadly Github doesn't allow me to upload more than 100 pictures)
This model is not giving a good accuracy due to the lack of data and also, due to the lack of a more sophisticated model. Probably CNNs will perform better than normal neural neworks.