Qwen3-VL flags a burned-in clinic name that another PII model missed
MaziyarPanahi · x · 2026-07-22
Qwen3-VL catches a clinic name that another PII model missed
A de-identification pipeline was tested on a chest X-ray and the results disagreed:
- OpenMed's PII model said the image was clean.
- Qwen3-VL re-read the pixels and refused to approve it, correctly spotting a clinic name still burned into the image.
- The author then ran a second round, and none survived after the follow-up check.
The post ends with a practical question: should every scan get a second reader? Both models are open on Hugging Face.
More from Multimodal
- An experiment asks whether a Midjourney image can become a style guide — ColleenMBrady · 2026-07-22
- Kimi K3 uses Unity CLI to generate a playable flying-dragon scene — bdsqlsz · 2026-07-22
- SenseNova-U1-Infographic-V3 adds real image editing for infographics — No-Measurement-5858 · 2026-07-22
- Qwen-Image-3.0 passes a hands-on test on Chinese text, six-language layouts, and image editing — 卡尔的AI沃茨 · 2026-07-22
- DJI Avata 360 footage turned into a 3DGS reconstruction of a Japanese dam — janusch_patas · 2026-07-22
- Miora tops Product Hunt with a memory-based creative studio for image, video, text and 3D — thisguyknowsai · 2026-07-22