research area
Computer Vision
We study recognition, detection, and segmentation under real-world conditions — distribution shift, long-tailed categories, and multi-modal input. Our evaluation work asks not just whether a model is accurate, but where and why it fails, and builds that answer back into training.
Projects
computer-vision
computer-vision
ShiftBench
A benchmark suite for measuring detection and segmentation performance under real-world distribution shift.
Publications
How Well Do Concepts Transfer Across Multi-Modal Tasks?
Yuki Tanabe, Mira Solheim · NeurIPS · 2026
A Benchmark for Failure-Mode Detection Under Distribution Shift
Dev Anand, Mira Solheim · ICCV · 2025
Adversarial Robustness in Multi-Object Detection
Priya Natarajan, Mira Solheim · CVPR · 2024
Beyond mIoU: Real-World Evaluation for Segmentation
Dev Anand, Yuki Tanabe · ECCV · 2024