search

sjtuplayer / anomalydiffusion

[AAAI 2024] AnomalyDiffusion: Few-Shot Anomaly Image Generation with Diffusion Model
MIT License
114 stars 14 forks source link

Could you tell me how to view the images generated by the model? #12

Closed boxbox2 closed 4 months ago

boxbox2 commented 4 months ago

In the "test_unet.py" , I only see the generation of masks, but I haven't found the generation of anomaly images. Can you tell me how to view the generated data?

sjtuplayer commented 4 months ago

You can just download the generated data from the google drive. And "test_unet.py" is to test the anomaly detection accuracy rather than generating the anomaly images. If you want to generate anomaly data yourself, please wait for the inference code, which will be released this week.