Skip to content
Implement of 'Beyond Frontal Faces: Improving Person Recognition Using Multiple Cues' by Ning Zhang, et al
Branch: master
Clone or download
sciencefans Global model and head-only model is implemented.
Global model and head-only model is implemented.
Using SVM with libnear, global model only get accuracy of 32.4% and
head-only model get accuracy of 78%.
After that, I use 1-NN to find the most nearest cosine distance
neighbor. This time, head-only model get 91%.
Latest commit 99bda0a Aug 27, 2015
You can’t perform that action at this time.