M. Usman Rafique
Research Focus
I am a researcher specializing in computer vision and machine learning. My primary areas of interest include image synthesis, image segmentation and understanding, and perception for robotics. My work has involved diverse settings, including natural outdoor scenes, remote sensing, and computer vision for logistics and warehouse robots.
Recently, I’ve been exploring large language models (LLMs) such as GPTs. Notable projects include:
- GPT-Nano: A lightweight implementation of GPT
- LLM-Forge: A playground for building practical LLMs with limited compute resources
Academic Background
I completed my PhD at the University of Kentucky, where I was a member of the Multimodal Vision Research Lab. My research focused on combining information from multiple images for scene understanding and image synthesis. My PhD advisors were Dr. Nathan Jacobs and Dr. Samson Cheung
Professional Experience
- Current Position (Since Aug 2023): Senior Machine Learning Engineer at Bastian Solutions RnD
- Developing computer vision and machine learning solutions for autonomous robotic systems in warehouses and logistics.
- Aug 2021 - May 2023: Senior Research and Development Engineer at Kitware Inc.
- Worked on large-scale change detection from overhead images and person identification and image restoration.
Old Teaching Website
My old website, from my teaching days is available here.
Research Projects
Near-Remote Sensing
Diverse View Synthesis
Novel View Synthesis
Multi-Image Fusion
Weakly Supervised Segmentation
Recent News
- Sep 17, 2024: released LLM-Forge Library on Github, a playground for building and training practical LLMs with limited computational resources.
- May 23, 2024: recognized as an outstanding reviewer for CVPR 2024. Pleased to be among top 2% of 9872 reviewers.
- Aug 21, 2023: glad to join RnD Lab of Bastian Solutions (a Toyota Advanced Logistics Company) as a Senior Machine Learning Engineer. I will be doing research on computer vision models for robotics and automation.
- June 5, 2023: implemented GPT-Nano, a light-weight large language model (LLM), implemented from scratch in PyTorch.
- May 6, 2023: recognized as an outstanding reviewer for CVPR 2023. I am very pleased to be one of 232 outstanding reviewers out of a total of 7000 reviewers.
- Jan 20, 2023: two papers accepted at IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2023.
- Jan 20, 2023: wrote a blog post: “Reflections on Reviewing Computer Vision Papers”.
- Aug 17, 2022: paper “Handling Image and Label Resolution Mismatch in Remote Sensing” (PDF) accepted to WACV 2023.
- March 3, 2022: paper “Revisiting Near/Remote Sensing With Geospatial Attention” (PDF) accepted to CVPR 2022.
- Feb 15, 2022: paper on sinkhole segmentation published to AGU Earth and Space Science Journal
- Nov 24 2021: recognized as an outstanding reviewer for BMVC 2021.
- Aug 2, 2021: joined Kitware Inc. as a Senior Research and Development Engineer
- June 8, 2021: I have successfully defended my PhD dissertation :confetti_ball: Bonus: the announcement tweet by my advisor
- May 20, 2021: recognized as an outstanding reviewer for CVPR 2021
- April 11, 2021: paper acceptd to NTIRE: New Trends in Image Restoration and Enhancement workshop and challenges at CVPR 2021
- March 16, 2021: paper accepted to IEEE International Geosciences and Remote Sensing Symposium (IGARSS) 2021
- December 12, 2020: gave a talk on “Automatic Identification of Sinkholes Using Deep Learning from Remote Sensing Data” at Kentucky Geological Survey
- July 31, 2020: paper accepted to BioImage Computing (BIC) workshop held at ECCV 2020
- July 29, 2020: paper accpeted to The British Machine Vision Conference (BMVC) 2020
- March 29, 2020: paper accepted to IEEE International Geosciences and Remote Sensing Symposium (IGARSS) 2020
- December 10, 2019: successfully defended my dissertation proposal
- June 17, 2019: presented my paper at EarthVision 2019 (CVPR 2019), Long Beach, CA
- April 5, 2019: paper accepted to IEEE International Geosciences and Remote Sensing Symposium (IGARSS) 2019
- April 4, 2019: paper accepted at EarthVision Workshop 2019 held in conjunction with CVPR 2019