Challenges

The customer was looking to build an AI based application to track and avoid repeat exams from the same students.

Following were the requirements from the customer

  • Validate Passport size photo upload with some resolution criteria.
  • Capture and match real time face detection with passport size photos of the same person stored in the system.
  • 60% Face coverage in the photo
  • Match photo stored with RealTime photo when candidate appears for the exam.

Variance Team’s AI base solutions to the Customer

  • Identify the orientation of the image whether is portrait or landscape.
  • The team did an implementation using pillow tools to reject those images which have a landscape view.
image-orientation

Support multiple image [JPEG, JPG, PNG] format when uploading

  • VIPL team implements a solution for image quality validation.
  • Different types of checks have been implemented as below,

    Blur

    image-quality-check-blur

    Bad Contrast

    image-quality-bad-contrast

    Glare/Over-Exposure

    image-quality-check-over-explosure
  • With the above checke it will pass for further analysis of facial recognition.
  • Variance Team use a technique to validate Face validation to verify that the image contains the face of the person at the proper angle.
  • Also, the face of the person should cover some percentage of the image e.g.: there was a need that the face must cover 60% of the image else the image is of no use.
face-validation
  • Image comparison technique for fraud detection in examination centers.
  • Variance AI team implemented image comparison with stored Photo during registration with the real-time photo of the student.
  • As per the criteria set it will allow/reject students to appear for the examination if the image is mismatched.
  • image-comparision image-comparision
  • Image comparison also validates Background Verification with color based on RGB values of the image.
  • Image resizing technique is used to resize images based on aspect ratio.
  • Variance AI team used the following scenarios to manage image resizing.
    • From aspect ratio and height, it will calculate the width
    • From height, it will calculate the width and aspect ratio and resize the image accordingly.
    • From height and width, it will auto-resize the image accordingly.
    • image-resizing
  • With image resizing, it will be easy for the exam center to properly validate face detection.

This feature was developed by the Variance AI team to compress images using various algorithms without loss the image quality.

Lossless-image-comparision

The following technique has been used to compress images.

  • RLE (Run-length encoding)

Outcome

  • With facial recognition systems it reduces chances of the fake examination by comparing photos stored in the system with actual face during examination.
  • Importing an image with an optical scanner or digital photography.
  • Analysis and image management including data compression and image enhancement and visual detection patterns such as satellite imagery.
  • It produces the final stage where the result can be changed to an image or report based on image analysis.

Technology we used

Python, OpenCV, Scikit-Image, Albumentations, FastAPI, MongoDB

python
opencv
scikit
albumentations
fastapi
mongodb

Trusted Customers

Contact Us About

By sending this form I confirm that I have read and accept Variance Infotech Privacy Policy

what happens next?

  • Our sales manager reaches you out within a few days after analyzing your business requirements
  • Meanwhile, we sign an NDA to ensure the highest privacy level
  • Our pre-sale manager presents project estimates and approximate timeline

We use cookies to provide better experience on our website. By continuing to use our site, you accept our Cookies and Privacy Policy.

Accept