Already available solutions for document verification

[afrozchakure]

Possible Solutions to Document OCR with links:

1. Adhaar Card OCR using Tesseract Library
Link to repo: Here

2. Pan Card OCR using Tesseract Library
Link to repo: Here

3. Tesseract OCR Pyimagesearch
Link to article: Here

4. OpenCv OCR + Tesseract with LSTM
Link to article: Here

5. Using Computer Vision API services (for higher accuracy and if the system will have an internet connection):

• Google Vision API OCR Engine
• Amazon Rekognition
• Microsoft Cognitive Services

6. Tesseract + OpenCV + Google API
Link to repo: Here

OCR has two major building blocks:

1. Text Detection - Detecting the required text is a tough task but thanks to deep learning, we’ll be able to selectively read text from an image.
2. Text Recognition - Recognizing and returning the text in the image.

Text Detection 2 approaches:

1. Region-Based Detectors
   • The first objective is to find all the regions which have the objects and then pass those regions to a classifier, which gives us the locations of the required objects. So, it is a two-step process.
   • Algorithms like Faster R-CNN and R-FCN take this approach.
   • This approach is considered more accurate but is comparatively slow as compared to the Single Shot approach.
2. Single Shot Detectors
   • Single Shot detectors, however, predict both the boundary box and the class at the same time. Being a single step process, it is much faster.
   • Single Shot detectors perform badly while detecting smaller objects.
   • SSD and YOLO are Single Shot detectors.

Important approaches:

1. Yolov3 Architecture
2. Faster RCNN Architecture

Common Frameworks being used:

1. Tesseract
2. OpenCV
3. Numpy
4. Tensorflow