You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Recently we added OpenCV to Pyodide. While testing OpenCV, I was able to run Pytorch neural network (converted to ONNX) in Pyodide using OpenCV DNN modules, though it is inference only and of course very slow (no GPU).
I would like to share some ideas about it, which also include some examples of handling images and dealing with the file system.
Very nice demo, looking forward to reading your post!
On the demo page, you might want to put a sentence mentioning OpenCV and ONNX (or put it in the title). And also enable re-sizing by default, otherwise it's will take a long time if the user uploads some high res image.
Personal opinion, and feel free to ignore, but it would be nice if the post also answered the question:
There is opencv-wasm and ONNX which work already in the browser, so why do we need Pyodide to do this?
Recently we added OpenCV to Pyodide. While testing OpenCV, I was able to run Pytorch neural network (converted to ONNX) in Pyodide using OpenCV DNN modules, though it is inference only and of course very slow (no GPU).
I would like to share some ideas about it, which also include some examples of handling images and dealing with the file system.
You can test it here: https://ryanking13.github.io/style-transfer-pyodide/
Demo video
demo.mp4
The text was updated successfully, but these errors were encountered: