Text extraction with OCR

Using EasyOCR our project can take in an image file, detect text in English, process, and output the text as a string along with the coordinates of each character.

Comment

Inspiration

Our main source of inspiration was finding similar projects online on websites such as stack overflow or GitHub and we found 2 main programs to try in our code which is Pytesseract and EasyOCR.

What it does

What our program does is first installs EasyOCR, then imports EasyOCR, MatPlotLib, cv2, etc. Next, it takes in the path to an image of your choice, displays it, then proceeds to output all detected English text line by line with the coordinates of the line, and the confidence that the text is correct.

How we built it

We built it by doing research on EasyOCR, how to use it, what methods to use, and what it outputs. After that, we were able to import it into our file, run the correct method with the image path, and successfully have some interpreted text.

Challenges we ran into

The main challenge that we ran into was that when you use an application like that it's not completely accurate for the font that was given on the comic strips and the only way we would be able to fix it is if we change the source code of EasyOCR which would be very difficult.

Accomplishments that we're proud of

Although we may not have gotten the interpretation to be as accurate as we would've liked we are proud that we were at least able to accomplish what we did when we came to this competition with almost no knowledge of image processing or machine learning.

What we learned

We learned how to process images and do different things with them through python along with different applications we can use in python and how to import and use them which expanded our knowledge immensely from what it was when we came into this competition.

What's next for Text extraction with OCR

Next for text extraction with OCR would be doing more research into EasyOCR and either finding a way to edit the code to be able to interpret the font we were given better or using that code to start a new application from scratch for OCR to do exactly what we want.

Built With

  • colab
  • cv2
  • easyocr
  • matplotlib
  • python
Share this project:

Updates