- Manan Chawla
- Anishka Gupta
AutoEval Challenge
The goal of the AutoEval Challenge is to develop a system capable of detecting handwritten true/false answers from scanned student PDF documents and calculating the corresponding scores automatically. This challenge requires the application of computer vision techniques and machine learning algorithms to accurately interpret handwritten responses and evaluate them.
https://huggingface.co/OpenGVLab/Mini-InternVL-Chat-4B-V1-5
This repository contains all the necessary files to reproduce our solution for the AutoEval Challenge. The main components of our solution include:
- Jupyter Notebook (
AutoEval_Challenge_Solution.ipynb): A well-documented notebook that walks through the entire process of our solution, including data preprocessing, model training, and evaluation. - split.py: A file while splits the images based on image mapping in different folders.
- join.py: A file while joins the csv output files.
To run the code in this repository, you need to have the following dependencies installed:
- Python 3.10 or higher
- Jupyter Notebook
- flash_attn
- Transformers
- PyTorch
- Pandas
For any questions or inquiries, please contact:
- Manan Chawla: mananapeejay@gmail.com
- Anishka Gupta: anishkagupta25@gmail.com
We hope this repository helps you in understanding and reproducing our solution for the AutoEval Challenge. Thank you for your interest in our work!