Special Session 4

Special Session 4: Generative AI for Image Reconstruction and Anomaly Detection

Description: Recent advances in generative artificial intelligence—particularly Generative Adversarial Networks (GANs) and Diffusion Models—have significantly pushed the boundaries of imaging science. These models exhibit superior capability in handling noise, reconstructing high-quality images, detecting anomalies, and integrating imaging physics into data-driven frameworks. By learning complex data distributions, generative models can produce realistic estimates of clean images, high-resolution outputs, and normal structures. Such capabilities enable more robust and stable performance than traditional convex or iterative optimization methods, which typically aim to find a single deterministic solution.

This Special Session seeks to bring together researchers working on cutting-edge generative techniques and their applications across image reconstruction and anomaly detection. We invite contributions that explore algorithmic innovations, theoretical insights, interpretable model, and real-world applications across medical imaging, remote sensing, auto-driven and related domains.

Session organizers
Assoc. Prof. Min Li, Hohai University, China
Assoc. Prof. Yibin Tang, Hohai University, China

The topics of interest include, but are not limited to:
1. Generative Models for Image Reconstruction
• GAN-based or diffusion-based image denoising, deblurring, and inpainting
• Super-resolution reconstruction using generative priors
• Generative models for multimodal or cross-domain image reconstruction
2. Anomaly Detection
• Reconstruction-based anomaly detection with GANs/diffusion models
• Latent-space generative modeling for anomaly identification
• Anomaly detection using generative likelihoods or score-based methods
3. Hybrid and Physics-Guided Generative Models
• Combining generative priors with traditional optimization
• Plug-and-play priors and score-based inverse problem solvers
• Learned data-consistency modules and reconstruction pipelines
4. Applications Across Domains
• Medical imaging diagnostics and biomarker extraction
• Remote sensing image analysis and enhancement
• Industrial defect detection and quality inspection

Submission method
Submit your Full Paper (no less than 8 pages) or your paper abstract-without publication (200-400 words) via Online Submission System, then choose Special Session 4 (Generative AI for Image Reconstruction and Anomaly Detection)
Template Download

Introduction of session organizers