Hi streamlit community!
I’ve made an anomaly detection app and there are three methods available: SPADE, PaDiM and PatchCore: https://share.streamlit.io/rvorias/ind_knn_ad
Developing the app felt as the cherry on top for my github repo. I’m surprised with how fast you can make a cool app showcasing some functionality!
There are some minor bugs such as image flickering and my wget std.out not showing up during dataset downloads. I’m also not sure if a GPU is used in the back-end. However, the algorithms are running fine.
2 Likes
That looks great @Raphael_Vorias!
I also really like the ability to custom upload your own data too! 
Best,
Charly
@Raphael_Vorias I tried to access your app but seems like he isn’t working anymore…
Thanks @Charly_Wargnier & @feliperoque for your feedback!
I’ve reduced the datasets and available models, plus fixed a PatchCore bug. I hope the app is a bit more stable now. If you guys have any more tips, I’d love to hear them!
Cheers!
@Raphael_Vorias Hi, if you try to access the link you made available above, he still doesn’t working…
@feliperoque Refreshing does not work?
I already use the st.session_state for both the data and the model.
I hope this works the same as st.cache 
Hmm, I’m running a bit out of options.
If you really want to try the app, you can also run it locally! GitHub - rvorias/ind_knn_ad: Industrial knn-based anomaly detection for images. Visit streamlit link t
I checked again the link of your app and works fine!