Need help with hosting a vton model on serverless
I'm trying to run the CatVTON model (https://github.com/Zheng-Chong/CatVTON, https://huggingface.co/spaces/zhengchong/CatVTON) using RunPod serverless. (Any vton model is fine as long as I can get the output)
My Goal:
A Flutter app sends an image to a Flask backend.
Flask forwards the image to a RunPod serverless instance running the model.
The processed image is sent back to Flask, then returned to Flutter.
I need a beginner-friendly guide to set this up. It's just for demo purposes, so I prefer serverless to save costs. I previously tried Google Colab Pro but ran into too many errors when trying out AI models.
I’m a bit scared about increasing costs while doing and testing this myself, and I worry I might end up paying a large amount.
Would appreciate insights on Hugging Face, RunPod serverless, and setting up this pipeline properly!
6 Replies
You can see some runpod docs to get familiar then continue on seeing hf code n docs
Or maybe the code for that specific app
https://discord.com/channels/912829806415085598/1349683836945960960/1349683949944963113
I got this reply from the bot
Is this correct?
If it is I can go ahead without worrying about costs
No it's a wrong method, I runpod serverless you gotta make a runpod specific worker handler, I guess it's not in that img from hfspaces
Yeah as long as you follow runpod docs, understand things well, I guess it's okay, if you wanna make sure something in your process just ask
It's better than having a full guide, because it's rare to have a full guide for a specific use case
https://github.com/omeshapasan2/CatVTON-Runpod-Serverless
anyway I made this
And there is some issues
GitHub
GitHub - omeshapasan2/CatVTON-Runpod-Serverless
Contribute to omeshapasan2/CatVTON-Runpod-Serverless development by creating an account on GitHub.
i see there's some exceptions there, i think its important to fix it first