Hey there, fellow machine learning enthusiasts! I recently stumbled upon the IDM-VTON project on GitHub, which uses Stable Diffusion to create realistic virtual try-on experiences. I was excited to dive in and understand how it works, but I hit a roadblock – memory issues.
I followed the README instructions on my laptop, but it wasn’t usable due to hardware limitations. Then, I tried cloning it to Google Colab, which initially gave me dependency issues that I solved using Miniconda. However, when I ran the `gradio_demo/app.py` script, the process got killed due to out-of-memory errors.
I’m now on the hunt for suggestions on running this project without a local GPU. Has anyone successfully run IDM-VTON or a similar virtual try-on model without a powerful GPU? I’m open to free or paid solutions as long as they let me test and understand the code.
If you have any tricks for optimizing memory usage in Colab or know of alternative tools or platforms that can help me achieve my goal, please share! I just want to be able to run this project, test it, play with the code, and see the results.
Let’s explore some possible solutions together.