Niki Amini-Naieni
I am a computer vision researcher working on visual understanding of the open world, with a focus on object counting, grounding, multi-task learning, and robust perception across images and videos.
Before joining New College, I am completing a six-month research internship at Google DeepMind. My work aims to build models that reason more flexibly about complex visual scenes and that can serve as useful tools for science, engineering, and real-world decision-making.
Google Scholar · GitHub · LinkedIn · CV
About
My long-term goal is to build general-purpose systems for visual understanding that can work reliably in open-world settings, especially where traditional object detectors and closed-vocabulary models fall short. I am particularly interested in methods that combine strong representation learning with grounded, controllable outputs.
At New College, I will pursue a five-year research programme while contributing tutorial teaching in Computer Science and participating in the College’s academic community. I am also interested in collaborations across Oxford, especially in domains where better counting and visual reasoning tools could accelerate research.
Research
Research themes
- Open-world object counting in images and videos
- Grounded visual reasoning and controllable prompting
- Multi-task visual understanding models
- Robust perception for scientific and real-world applications
Current focus
Use this box for a short description of your current fellowship agenda. For example: building a unified model for counting, grounding, localization, and description across diverse visual settings.
Selected Publications
Teaching
Add your tutorial teaching, invited teaching, guest lectures, course leadership, and mentoring here.
- Advanced Deep Learning, University of Oxford — Co-Lead
- Undergraduate tutorials in Computer Science — New College, Oxford
- Guest lectures / workshops / mentoring
Software & Datasets
- CountVid — dataset / benchmark for open-world object counting in videos.
- Project Name — short description and link.
- Project Name — short description and link.
Talks
Add selected recorded talks, invited talks, and conference presentations. You can embed YouTube videos directly here.
Supervision & Collaborations
Use this section to highlight students you supervise, mentees, and key collaborators. This is especially important for demonstrating your ability to build and lead research directions.
- PhD / MSc students (names, topics, institutions)
- Undergraduate supervision and mentoring
- Collaborators across academia and industry
CV
Upload your CV as cv.pdf and link it here. You can also add short links to your full publications list,
awards, talks, and service.
Contact
your.name@new.ox.ac.uk
New College, University of Oxford
Google Scholar · GitHub · LinkedIn
Add a short note about collaboration inquiries here.