Projects I've worked on over the years.

I’ve worked on tons of research projects over the years. Many of them are open-source, so if you see something that piques your interest, check it out.

  • Atria Core is a PyTorch-based toolkit designed for training and testing machine learning and deep learning models at scale.
    github.com
  • TorchXAI provides efficient implementations of several XAI (Explainable AI) metrics that seamlessly integrate with the Captum library.
    github.com
  • An approach to generate differentially private document image datasets using layout-conditioned latent diffusion models
    github.com
  • An approach for augmenting existing CNN-based classifiers with pyramid-attention modules to introduce inherent interpretability.
    github.com
  • An approach to document image binarization through cold diffusion.
    github.com
  • A complete toolkit for active learning for deep learning classification tasks.
    github.com
  • Toolkit for generating and evaluating corruptness robustness benchmark for document images.
    github.com
  • Toolkit for explainable AI for document image classification.
    github.com
  • A complete toolkit for integrating MoveIt motion planning with MAVROS-based aerial robots in ROS.
    github.com
  • A toolkit for integrating Gym with ROS for training reinforcement learning agents on robots in ROS/AirSim-based simulated environments.
    github.com
  • A NVIDIA Deepstream pipeline for real-time detection and tracking of individuals in media news channels.
    github.com
  • A NVIDIA Deepstream pipeline for real-time tracking of traffic and analytics in retail stores.
    github.com
  • AutoFortnite project was about controlling a fortnite player with external controllerinputs autonomously with realtime streamed visual game inputs handled by NVIDIA DeepStream SDK.
    github.com
  • A ros based navigation package for 2d mobile robots (focused on Pioneer 3AT) developed at IRL Robotics Lab, NUST.
    github.com
  • The codebase for Team NUST's Standard Platform League (SPL) team for RoboCup, using NAO humanoid robots.
    github.com
  • A frontier based exploration algorithm for navigating through unknown environments and generating a map based on octomaps.
    github.com
  • A custom ROS-based implementation of the popular ORB-SLAM algorithm for monocular/RGB-D cameras with bundle-adjustment.
    github.com
  • An octomap based VFH+ algorithm in 3D for local planning of UAVs (or any robot in 3D).
    github.com