Head pose estimation network based on simple, handmade CNN architecture. Angle regression layers are convolutions + ReLU + batch norm + fully connected with one output.
|Supported ranges||YAW [-90,90], PITCH [-70,70], ROLL [-70,70]|
|Angle||Mean ± standard deviation of absolute error|
|yaw||5.4 ± 4.4|
|pitch||5.5 ± 5.3|
|roll||4.6 ± 5.6|
1, 3, 60, 60 in
1, C, H, W format, where:
C- number of channels
H- image height
W- image width
Expected color order is
Each output contains one float value that represents value in Tait-Bryan angles (yaw, pitch or roll).
1, 1- Estimated yaw (in degrees).
1, 1- Estimated pitch (in degrees).
1, 1- Estimated roll (in degrees).
[*] Other names and brands may be claimed as the property of others.