Niki is the first Governing Body Fellow in Computer Science at New College. Her research focuses on artificial intelligence, computer vision, and deep learning. She is best known for her work developing deep learning models for automatically counting and detecting any type of object in images and videos. During her DPhil at the University of Oxford, she developed methods for open-world object counting using natural language and visual prompts and extended these approaches from images to video. Prior to her doctoral studies, Niki worked in computer vision, applied mathematics, and robotics at SpaceX (Starlink) and Amazon (Project Kuiper, now Amazon Leo). She is currently applying her doctoral research to Gemini at Google DeepMind.

Teaching:

Niki teaches undergraduate tutorials in Computer Science at New College. She also teaches doctoral students as a Co-Lead and Lecturer on the Intelligent Earth Advanced Deep Learning Course and supports the Computer Vision and Deep Learning practicals for the Autonomous Intelligent Machines and Systems Course. Occasionally, she offers tutorials to Schmidt AI in Science Postdoctoral Fellows on applying computer vision and deep learning to the sciences.

Research Interests:

Niki's research focuses on computer vision, deep learning, and artificial intelligence, with a particular interest in building general-purpose systems for visual understanding. She develops models that can count, detect, segment, and track any specified type of object in images and videos using flexible combinations of natural language and visual examples.

A central goal of her research is to develop AI systems that can efficiently and accurately analyse scenes containing hundreds or thousands of objects without requiring task-specific training. These systems can produce object locations and segmentation masks, track individual objects through video, and estimate both frame-level counts and global counts of unique objects across a video. Her research is motivated by applications across biology, medicine, materials science, Earth observation, and the humanities.

More details can be found on her website.

Selected Publications:

• Niki Amini-Naieni and Andrew Zisserman. CountGD++: Generalized Prompting for Open-World Counting. CVPR, 2026.
• Niki Amini-Naieni and Andrew Zisserman. Open-World Object Counting in Videos. AAAI, 2026.
• Niki Amini-Naieni, Tengda Han, and Andrew Zisserman. CountGD: Multi-Modal Open-World Counting. NeurIPS, 2024.
• Niki Amini-Naieni, Kiana Amini-Naieni, Tengda Han, and Andrew Zisserman. Open-World Text-Specified Object Counting. BMVC, 2023.