Graph2Text: Get Textual data from Bar charts

Please refer academic poster.pdf for some additional details

Requirements

• PIL==10.0.1
• matplotlib==3.7.2
• pandas==2.1.1
• pytesseract==0.3.10
• sklearn==1.3.2
• torch==2.1.0
• torchvision==0.16.0
• tqdm==4.66.1

In addition to these python libraries, please make sure to download Tesseract OCR in your machine as pytesseract is just a wrapper that calls Tesseract OCR. Download Tesseract from https://tesseract-ocr.github.io/tessdoc/Downloads.html.

Code

The code folder contains two jupyter notebooks:-

• Graph2Text = This is the main system notebook which all training and predicting code
• Graph2Text_Predict_one_Image = This is a notebook which only has prediction code. It takes the path of a single image and predicts output for that image.

To use Graph2Text

To use the Graph2Text notebook, please follow the following steps:-

• Set images_path to path of images folder
• Set annotations_path to path of json annotation folder
• Set model_path_loss to where you would like to store and extract the model from
• Set pytesseract.pytesseract.tesseract_cmd to the absolute path of tesseract executable
• Run the notebook

To use Graph2Text_Predict_one_Image

To use the Graph2Text notebook, please follow the following steps:-

• Set image_path to path of image
• Set model_path_loss to where you would like to store and extract the model from
• Set pytesseract.pytesseract.tesseract_cmd to the absolute path of tesseract executable
• Run the notebook