Model input RGB or BGR?

[yataoz]

Hello,

I'm trying to figure the correct input channel order for the model. In rt_gene_standalone/estimate_gaze_standalone.py, you used cv2.imread to read the image as BGR, and I don't see anywhere else in the following steps to convert it to RGB. However, in your trainining code, seems like you used RGB in rt_gene_model_training/pytorch/rtgene_dataset.py . Which one is the correct order? Thanks!