Abdul Rahman Email & Phone Number

Tokyo, Japan

Islāmābād, Pakistan

National Science & Technology Park, National Univeristy Of Sciences And Technology, Islamabad

Developed semantic segmentation models for mobile devices. - Developed light version of UNet using Separable Convolutions. - Trained and deployed models for clothes segmentation. - Trained and deployed models for facial features segmentation.

Developed general OCR model based on CRNN architecture - Trained model for Urdu Synthetic text recognition - Deployed model to web for testing - Published paper in ASAR 2021 workshop (achieves better accuracy in Urdu and Arabic text recognition than Google Vision APIs)Trained EAST Text detection model for Urdu News Channels - Developed Web-based data.

Germany

Worked on two separate projects during my summer internship in DFKI.1) DeepCFD: Efficient Steady-State Laminar Flow Approximation with Deep Convolutional Neural Networks - Computational Fluid Dynamics (CFD) simulations are computationally expensive and we tried to approximate velocity and pressure of fluid if an object of specific geometry is placed at.

Worked on document analysis for extracting information from legal documents.

Pakistan

Developed AI system for semantic video segmentation - Clay Pigeon segmentation in video streams - Segmentation on CamVid dataset

Bachelor Of Engineering - Be, Computer Software Engineering

Education record