INFO.txt

# Team Creatorz

## Model Info

1. We have trained our model on 30 characters which can be found in our character set.
2. We trained our model on our custom dataset. The dataset have been made using letters of different width, rotations, noise etc.
3. The model can predict characters with different width. Letters written by pens, sketch pens can be easily detected. (Do not use markers).

## Image Segmentation Info

1. Our segmentar can deal with a rotation range of -45 to 45 degrees majorly. In some cases it can also deal with a rotation range of -60 to 60 degrees.
2. Our segmentar can also remove shadow upto certain limit.
3. Our segmentar can detect small line breaks between a word.

## Testing

1. Install the dependencies using the requirement.txt file.
2. Run the main.py file in suitable environment
3. When it will ask for an image path give it a correct image path (If the image path is not correct it will again ask for an image path again)
4. The output will show the segmented characters and it will print the predicted letters.
5. To stop the run press escape button. 

## Character Set

"क", "ग", "घ", "च", "छ", "ज", "झ", "ञ", "ट", "ठ", "ड", "ण", "त", "थ", "द", "न", "फ", "म", "र", "ल", "व", "श", "त्र", "ज्ञ", "अ", "उ", "ऊ", "ए", "ऐ", "ओ"