Official implementation of the paper "Estimation of continuous valence and arousal levels from faces in naturalistic conditions", Antoine Toisoul, Jean Kossaifi, Adrian Bulat, Georgios Tzimiropoulos and Maja Pantic, Nature Machine Intelligence, 2021
I put the image and bounding box in the way you implemented in Data Augmentor and converted Affine.
However, the value of VA comes out in units of 1000 and the emotion comes out weird,
When I divided the Affine transformed value by 1/255, I started to get some value.
I'm wondering if this conversion method is correct, and if possible, I'd like to see the preprocessing method I showed in the demo gif.
I want to use your great model.
I put the image and bounding box in the way you implemented in Data Augmentor and converted Affine.
However, the value of VA comes out in units of 1000 and the emotion comes out weird, When I divided the Affine transformed value by 1/255, I started to get some value.
I'm wondering if this conversion method is correct, and if possible, I'd like to see the preprocessing method I showed in the demo gif.
Thank you.