Image Quality Scanner
Our project determines the quality of the documents before they are processed through OCR and other processing methods.
Created on 12th February 2023
•
Image Quality Scanner
Our project determines the quality of the documents before they are processed through OCR and other processing methods.
The problem Image Quality Scanner solves
In the ever-evolving metaverse, the need for digitising text documents and
extracting their essential data is becoming more crucial. However, the quality of
these documents can often be poor, resulting in low-quality Optical Character
Recognition (OCR) data and poor document intelligence extraction. The solution to
this is to automatically determine the quality of these documents before they are
processed through OCR and other processing methods. The goal is to assess the quality of the documents and determine if they are fit to undergo OCR and further processing using computer vision, image
processing methods, and machine learning models, or any other solution
that provides the best outcome.
Challenges we ran into
Our project uses CNN to train and validate the dataset and detect the quality of documents by testing it. We were facing issues while implementing CNN and working with the Dataset as it had only 70 images. So we had to perform Data Augmentation on te data. Also we had to train the data a lot as we wanted to achieve 100 % accuracy and within a short time. We are proud enough that our algorithm was able to detect tge qulity of documents within less than a second.
Tracks Applied (1)