[CfP] VAND 4.0 Challenge @ CVPR 2026
Sebastian Höfer
Dear Community,
We are excited to announce the Call for Participation for the Visual Anomaly and Novelty Detection (VAND) 4.0 Challenge, part of the VAND 4th Edition Workshop at CVPR 2026!
We invite the global community of innovators, researchers, and technology enthusiasts to engage with these challenges and contribute to advancing anomaly detection technologies in real-world scenarios.
When? The competition officially runs from April 1 - May 14, 2026.
Prizes: Total prizes worth ~$25,500 USD awarded for Best Performance, Best Efficiency, Best Zero-Shot/VLM models, and Best Paper!
What? Two separate tracks, Industrial (MVTec-AD2) and Retail (Kaputt2)
Challenge Website: https://sites.google.com/view/vand4-cvpr2026/challenge
—
Challenge Description The challenge addresses critical industrial needs for reliable anomaly detection under varying conditions and with limited data. Building on the encouraging results from previous years, this edition sets its sights even higher. It pushes the boundaries of robust and generalizable anomaly detection models for real-world use cases by including, for the first time, both industrial and retail logistics-focused competitions. Our goal is to bridge academic research with industrial requirements to develop solutions directly applicable to manufacturing, retail logistics, and beyond.
Track Descriptions Participants can choose to compete in a single track or enter both via two separate submissions. Note that each track has its own submission platform.
Industrial Track (MVTec AD 2): This track utilizes the MVTec AD 2 dataset. Its design allows for evaluating models under real-world distribution shifts induced by changes in lighting conditions.
Track details: https://drive.google.com/file/d/1O9JCLXAA1lr9x8rkB0MF-YUEr843WJDg/viewRetail Track (Kaputt 2): This track utilizes the Kaputt 2 dataset (an extension of the 2025 Kaputt dataset). It illustrates the scientific challenges for defect detection arising in retail logistics scenarios, such as heavy pose and appearance variations.
Track details: https://drive.google.com/file/d/1Tbrw-P8BsYtETm1EmeXlK8RdgZVGBpj1/view
Important Dates
Challenge Time: April 1st – May 14th (Anywhere on Earth - AOE)
Submission Deadline: May 14th, 23:59 PM AOE
Results Announcement: May 21st
Video Presentation Submission Deadline: May 28th
VAND 4.0 Workshop @ CVPR 2026: June 3rd/4th
Community Channel
Join the discussion, ask questions, and connect with other participants on our Discord channel:
Website: https://sites.google.com/view/vand4-cvpr2026/challenge
Email: vand4-0-...@googlegroups.com
Discord: #cvpr-challenge-vand4-0 (via the FiftyOne Community)
Organizing Team
MVTec: Lars Heckler-Kram, Jan-Hendrik Neudeck, Ulla Scheler
Intel: Samet Akcay, Ashwin Vaidya
Amazon: Sebastian Höfer, Dorian Henning, Anton Milan
Other: Paula Ramos
For full technical guidelines, rules, and to learn more about the dataset resources, please visit our official challenge website.
We look forward to seeing your innovative solutions – register now!
On behalf of the VAND 4.0 Organizing Team, Sebastian Höfer (Amazon, Fulfillment Technologies & Robotics)