HIGHLIGHTS
SUMMARY
The authors used the dataset and DogFACS annotations generated as part of a previous study by Bremhorst et_al22. The subjects were located behind a transparent window using the protocol which is fully described in Bremhorst et_al22. After a pre-processing step of extracting cropped dogs faces from the frames (see Fig 2 for examples), in the deep approach the raw cropped faces are taken as input by a neural_network. The authors experiment here with neural_network architectures of two types: convolutional neural_network (Resnet5056) and the recently introduced vision transformer57 (ViT) network. The set of the . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.