Charles Gaillard

Is it mandatory to support conda? If so, maybe we can switch to another pdf-reader lib.

So now with #829 we are just missing `weasyprint`, right @fg-mindee ?

I think we shouldn't only perform NMS, because here for instance we want to keep both boxes when there is an overlap. I see 2 solutions: - Merging the 2...

As a matter of fact, we do want to suppress very small boxes included in other ones, so I suggest the following: - performing NMS with a very high threshold...

**TF vs CV2** As mentioned [here](https://towardsdatascience.com/image-read-and-resize-with-opencv-tensorflow-and-pil-3e0f29b992be), cv2 uses by default bilinear interpolation with half-pixel corrections, while the bilinear resizing of tf doesn't use this half-pixel correction. The way to fix...

**PIL vs TORCH** [This issue](https://github.com/pytorch/vision/issues/2950) also tackles this issue of anti-aliasing between torch and PIL. As mentionned in [this article](https://tcapelle.github.io/pytorch/fastai/2021/02/26/image_resizing.html), the arg `anti_aliasing` is now available in torchvision: Torchvision.transforms.Resize: "The...

 This is quite a good summary, found [here](https://twitter.com/poolio/status/1280689329908543488?lang=en) it seems that using `anti_alias=True` in tf and `anti_aliasing` in torchvision lead to ALMOST similar results.

For the moment we used `tf.image.resize` and` torchvision.transforms.functional.resize` in all our preprocessors thus in all trainings (both reco & detection). We didn't use anti aliasing in TF nor in torch...

Yes, and also in the torch resizing function as mentioned [here](https://pytorch.org/vision/stable/transforms.html#torchvision.transforms.functional.resize): ``` antialias (bool, optional) – antialias flag. If img is PIL Image, the flag is ignored and anti-alias is...

Any updates on that @fg-mindee ?