Jia Guo
Jia Guo
Yes, test only.
Hi, try ``wget http://insightface.ai/files/models/arcface_r100_v1.zip``
insightface Python library supports face detection only, but it currently does not support face parsing (face masking). You can try using the dml_csr module: https://github.com/deepinsight/insightface/tree/master/parsing/dml_csr
You can do L2 distance loss mimic.
You can convert .pt model to .onnx using the dynamic shape on the first 'batch' axis.
Yes you only need to specify the locations of the 5 landmarks out of the 106 keypoints.
You can cite our github link directly that's no problem.
change to load the face-detection model under 'antelopev2' or 'buffalo_l' model package.
you can try training your own age classification model by using larger dataset.
1.0/0.0 is the flag of visible.