Open mxochicale opened 4 years ago
Hi Miguel,
I haven't done a formal study to assess the quality of detected landmarks in children, but subjectively the accuracy on children is a bit lower than that of adults (due to the data used to train OpenFace mainly consisting of adults, and children have fairly differently shaped faces)
Having said that, we have used OpenFace on datasets of children with moderate success: https://www.researchgate.net/profile/Behnaz_Nojavanasghari/publication/308407783_EmoReact_A_Multimodal_Approach_and_Dataset_for_Recognizing_Emotional_Responses_in_Children/links/59dbe1690f7e9b1460fc2c25/EmoReact-A-Multimodal-Approach-and-Dataset-for-Recognizing-Emotional-Responses-in-Children.pdf
Thanks, Tadas
Dear Tada
I am wondering if OpenFace has been used to identify face landmarks of Children. Also, I am assuming that children face landmarks can be identified with OpenFace but I am not sure about the validity of such localisation of landmarks. Any ideas?
Best wishes Miguel