Jaideep Randhawa
Machine learning researcher. I build models for bioenergy, power converters, swarm robotics and the eSanjeevani
national telemedicine service.
B.Tech in Computer Science and Engineering, Guru Nanak Dev Engineering College, Ludhiana. CGPA 9.33/10.
Download CV (PDF)
Every signature upload passes a gate.
At C-DAC Mohali I built a multi-stage validation pipeline for signature images uploaded to the eSanjeevani
national telemedicine service. It checks image integrity and quality, locates the signature with YOLOX,
classifies it with EfficientNet-B0 and rejects anything that is not a signature: logos, geometric patterns,
drawings, textures. The models were trained on more than 18,000 multilingual signatures (CEDAR, BHSig-Hindi,
BHSig-Bengali) plus more than 11,000 signature and non-signature samples.
Draw below. Your browser runs the quality and shape checks live. The trained models run on the server in the
real pipeline and are not simulated here. A scribble will pass the browser checks; telling a scribble from a
signature is the classifier’s job.
- Blank image
- Signature area
- Contour shape
- Integrity, resolution, blur and brightness checks
- Signature localization (YOLOX)
- Classification (EfficientNet-B0)
Let’s talk research.
Write to
jaideeprandhawa220@gmail.com.