Adel Basli

Hey, I am working on a similar issue (https://github.com/openfoodfacts/openfoodfacts-ai/issues/352). If we are interested about this issue, it might be interesting to start it in parallel. Let me know what you...

> In our tests filtering is working again on macOS 15.1 beta even with Firewall on. > > I'm not sure if fix is present in 15.0.1 released today. I...

This is the planned tasks, let me know if I am missing something - [ ] Test the fine tuned [CORD models](https://huggingface.co/naver-clova-ix) and see how they perform on Open Food...

As an update, I am still working on the adaptation of this fine tuned model: https://huggingface.co/nielsr/layoutlmv3-finetuned-funsd I was able to get the token identification working, but now I am looking...

Looking at the Document Intelligence Studio from Microsoft, it seems that their API is very accurate, and can take url as input to extract several key informations, each of them...

I can give it a try, if you can get me a list of a dozen of receipt that you want to benchmark, I can run the experiment and extract...

I would rather choose a simple outlier detection technique, as I am not sure we will have enough data to properly fit a model. IQR would be a solid fondation.

I am surprised to see that, what is the interest of having fake accounts approving PR in this repo ?

I was thinking we can label images with 3 labels: - product_and_price - product - price Basically we will draw a bog bounding box on product + price, then 2...

Basically, the labelling task would look like this: