ieee-vit / auto-grader Goto Github PK

Right now, one endpoint exists which takes in an image with the correct answers and returns the marks.
Create a new endpoint that will take in the student's answers with the correct answers in a JSON and return the marks.

Right now, the character recognition model works very well on the training data but fails to perform as good on the testing data.
Improve the accuracy of the whole OCR pipeline (extraction and preprocessing of the boxes and the classification model itself)

It also recommends installing indexer module, which is not compatible with python 3.
Originally posted by @amitpant14 in #3 (comment)

Create a Docker container (currently being worked on in the docker branch)

Fix the sorting of the boxes cropped so that it is sorted from top left to bottom right.

Document the endpoint functions to tell what the functions do.
The functions are grade and grade_text in main.py

The grade function takes in the image of the answer sheet, the correct answers, max marks and bias to calculate the marks to be given to a student.

The grade_text function takes in the student's answers in the form of text, the correct answers, max marks and bias to calculate the marks to be given to a student.

Things to do:

1. Add function comments to the above-mentioned functions in main.py.
   Example:

   def hello():
       """This a document string :)""" # This is how you write a doc string for a function
       print("HELLO")

2. Write about these two endpoints in the readme with sample input and output.

Right now, the answer sheets have to be perfectly aligned before putting them into the pipeline.
We have to automate that.

Right now, on hitting the base URL, the API just shows "Auto-Grader is online".
Create a better-looking homepage.