Method
Generative image models
Diffusion, GAN and inpainting models for image synthesis or restoration.
8 cases from 6 companies apply Generative image models. 1 was checked in an online A/B test.
Typical tasks: Evaluate models, Optimize decisions, Reduce ML cost, Label data with LLMs and 1 more. Companies: LY Corporation, Celsus, Etsy, Sber and 2 more. Often combined with: Data augmentation, LoRA fine-tuning, U-Net, Vision-language model, Image classification and 13 more.
Digital Twins of the Heart and Speeding Them Up![]() | ![]() | healthtech | 2026-05 |
Choosing an automatic quality metric for inpainting models by correlation with human ratings![]() | ![]() | social | 2025-06 |
Evaluating AI-generated images (Blackbox optimization)![]() | ![]() | general | 2025-06 |
Evaluating AI-generated images (The basics)![]() | ![]() | general | 2025-06 |
Improving Aerial Imagery Interpretation with Diffusion Networks![]() | ![]() | manufacturing | 2025-05 |
Diffusion Models for Synthesizing "Horrible" Medical Data![]() | ![]() | healthtech | 2025-05 |
| Efficient visual representation learning and evaluation | marketplace | 2024-06 | |
| Speeding Up PET/CT Scans with Deep Learning Denoising | healthtech | 2023-05 |



