VISION'26 @ CVPR 2026

Overview

Aims and Scope

The VISION workshop will provide a platform for the exchange of scholarly innovations and emerging practical challenges in Vision-based Industrial Inspection. Through a series of keynote talks, technical presentations, and challenge competitions, this workshop aims to (i) bring together researchers from the interdisciplinary research communities related to computer vision-based inspection; and (ii) connect researchers and industry practitioners to synergize recent research progress and current needs in industrial practice.

In 2026, VISION expands its scope to reflect the merger with the “DEgraded X-ray image Tomography, Enhancement, and Reconstruction” and “Intelligent Screening and Imaging For Detection and Exploitation via X-ray” workshops.

We explicitly welcome contributions across these three tracks, from traditional vision based methods to cross-modality methods and shared challenges in real-world inspection pipelines.

Topics include surface imaging, X-ray, CT, and volumetric reconstruction, 3D sensing modalities, computational imaging, imaging physics, multimodal cross-domain fusion, scene understanding for threat detection, datasets and benchmarking, and real-world deployment and applications.

Keynote Talks

Salman Khan

Computer Vision Department, MBZUAI

"Towards Generalist and Edge Intelligence for Industrial Inspection"

Dan Cristian Dinca

Rapiscan Systems

"Intelligent Security Screening (X-ray)"

Charles A. Bouman

Purdue University

"Past, Present, and Future Methods for Sparse View CT"

Paul Shahidi, PhD

Anduril Industries

"Why Most Industrial Vision Systems Fail their First Deployment."

Interested audience members may contact Dr. Shahidi directly at info@paulshahidi.com.

Amir Afrasiabi

The Boeing Company

"Self-Validating Augmented Reality Using Artificial Intelligence"

Jennifer Vandoni

Safran

"AI for Non-Destructive Testing and Material Characterization in Aerospace"

Interested audience members may contact Dr. Vandoni directly to request the slides at jennifer.vandoni@safrangroup.com.

Program

Agenda

Workshop slides are available in the VISION'26 slides folder.

Time	Topic
8:00 - 8:15	Chair's Opening Remarks
8:15 - 8:45	Invited Talk I: Paul Shahidi - Why Most Industrial Vision Systems Fail their First Deployment. Interested audience members may contact Dr. Shahidi directly at info@paulshahidi.com.
8:45 - 9:15	Invited Talk II: Jennifer Vandoni - AI for Non-Destructive Testing and Material Characterization in Aerospace Interested audience members may contact Dr. Vandoni directly to request the slides at jennifer.vandoni@safrangroup.com.
9:15 - 9:45	Invited Talk III: Dan Cristian Dinca - Intelligent Security Screening (X-ray)
9:45 - 10:55	Coffee Break & Poster Session
10:55 - 12:00	Flash Presentations AI-Driven Hyperspectral Imaging for Manufacturing Safety and Inspection Minjun Song EG-DINO: Guiding Foundation Models with Structural Priors for Domain-Incremental Crack Segmentation Bilal RAHOU FryNet: RGB-Thermal Fusion for Frying Oil Oxidation Taminul Islam Fine-grained Alignment for Counting Objects in Industry Suhye Cho Towards a Unified Framework for Transforming Vision Foundation Models into Zero-Shot Multimodal Anomaly Detectors Matic Fučka EviInspect: Evidence-Grounded Annotation and Evaluation for Safety-Critical Industrial Inspection Nikolaos Marios Militsis A Scale-Aware Industrial Vision Pipeline for Small Defect Inspection on Complex Gear Surfaces Yuan Wang Supervised Contrastive Learning for Industrial Defect Detection with Severe Class Imbalance Zhamilya Saparova RAGT-Net: Robust Adversarial Geometric Tension for Box-Supervised Industrial Defect Segmentation Zhiqian He
12:00 - 13:30	Lunch Break
13:30 - 14:00	Invited Talk IV: Charles A. Bouman - Past, Present, and Future Methods for Sparse View CT
14:00 - 14:40	Flash Presentations 3D Field of Junctions: A Noise-Robust, Training-Free Structural Prior for Low-Dose CT Narges Moeini- Namhoon Kim 3D Patch Diffusion Model for Volumetric Tomographic Reconstruction with View-Consistent Blind Deblurring Taewon Yang Physics Guided Threat De-occlusion/Restoration in Dual Energy X-Ray Baggage Images Krishan Sharma Fractional-gradient Sparsity with Autoencoding Sequential Deep Image Prior for 3D CT Reconstruction Chaoyan Huang Dataset and Baseline Method for Long-range X-ray Image Enhancement, 3D Reconstruction, and Object Recognition Riley O'Neill CT-DegradBench: A Physics-Informed Benchmark for CT Degradation Detection and Severity Estimation Taifour Yousra CryoGEN-II: Cryogenic Electron Tomography Reconstruction via Generative Network Yunfei Teng Self-supervised High-corruption Image Restoration for Low-dose CT Srijita Bandopadhyay
14:40 - 15:10	Invited Talk V: Salman Khan - Towards Generalist and Edge Intelligence for Industrial Inspection
15:10 - 15:40	Invited Talk VI: Amir Afrasiabi - Self-Validating Augmented Reality Using Artificial Intelligence
15:40 - 16:00	Coffee Break
16:00 - 17:00	Panel Discussion Panelists: Paul Shahidi (Anduril Industries), Dan Cristian Dinca (Rapiscan Systems), Charles A. Bouman (Purdue University), Salman Khan (MBZUAI), Amir Afrasiabi (The Boeing Company)
17:00 - 17:30	Flash Talks (paper presentation) Are Vision Foundation Models ready for Industrial Inspection? Mehdi Gharbage RadioCore: Few-Shot Industrial Anomaly Segmentation with Multi-Scale Radio ViT Features Mohammed Ahmed Mohammed Eltoum Mohammed Ali Anomaly-to-Normal Translation: A Training Paradigm for Feature-Based 3D Point Cloud Anomaly Detection Lars Fichtel Physics-Conditioned Video Diffusion for Nuclear Fusion Capsule Polishing Process Synthesis Shashank Galla GuidedFlow: An Attention-Guided Framework for Anomaly Detection in Additive Manufacturing Sosmita Paul Image segmentation and physics-guided regression for assessing laser surface structuring Vladimir Bukhtoiarov

Call for Papers

Topics of Interest (not limited to)

From an industry point of view:

New developments and evolving practices across industries such as automotive, electronics, and construction, focusing on materials like glass, carbon fiber, metal, and plastic.
Advances in vision-based inspection technologies, the rise of foundation models, and the challenges these bring in terms of algorithm development (e.g., multi-modal data fusion, physics-informed learning)
Integration of automated screening in airports, borders, and logistics
Mitigating evolving threat types, such as miniaturized explosives, dismantled improvised explosive devices
Edge and real-time implementations for high-throughput inspection and screening
Adaptive triage systems to optimize Human–AI collaboration
Imaging-based non-destructive testing
Release of large-scale industrial inspection and screening datasets.

From an academic point of view:

Algorithms for Data and Annotation Limitations
Techniques to Overcome Real-world Data Challenges
Data-centric Tools and Methodologies
Robust, Explainable, and Accountable Machine Learning
Meta-Learning, Transfer Learning, and Domain Adaptation
Foundation Models in Industrial Inspection and Security Screening
Multi-Modality, Multi-Energy Data Fusion for Inspection and Screening
Context-aware, fine-grained, and part-based analysis of concealed or dismantled threats
One-class and open-set recognition for rare or unseen threats
Generative AI and diffusion models for synthetic X-ray and CT data augmentation
Spectral and hyperspectral X-ray imaging for material discrimination
Physics-informed deep learning for material and substance classification
Quantitative imaging for distinguishing organic, inorganic, and metallic substances

Submission Instructions

We invite research and industrial papers relevant to the workshop topics. Authors may choose between:

Regular papers: Full-length submissions eligible for the workshop proceedings (details will follow the CVPR 2026 workshop guidelines).
Extended abstracts: Short submissions that will not appear in the proceedings.

Tracks (new for 2026)

This year, we offer three submission tracks. Authors are encouraged to submit to the track that best matches their paper's focus:

VISION'26 General Track
DEXTER Track: DEgraded X-ray image Tomography, Enhancement, and Reconstruction
INSIDE-X Track: Intelligent Screening and Imaging for Detection and Exploitation via X-ray