head-pose-estimation-adas-0001

Use Case and High-Level Description

Head pose estimation network based on simple, handmade CNN architecture. Angle regression layers are convolutions + ReLU + batch norm + fully connected with one output.

Validation Dataset

Biwi Kinect Head Pose Database

Example

head-pose-estimation-adas-0001.png

Specification

Metric Value
Supported ranges YAW [-90,90], PITCH [-70,70], ROLL [-70,70]
GFlops 0.105
MParams 1.911
Source framework Caffe*

Accuracy

Angle Mean ± standard deviation of absolute error
yaw 5.4 ± 4.4
pitch 5.5 ± 5.3
roll 4.6 ± 5.6

Performance

Inputs

  1. name: "data" , shape: [1x3x60x60] - An input image in [1xCxHxW] format. Expected color order is BGR.

Outputs

Output layer names in Inference Engine format:

  1. name: "angle_y_fc", shape: [1, 1] - Estimated yaw (in degrees).
  2. name: "angle_p_fc", shape: [1, 1] - Estimated pitch (in degrees).
  3. name: "angle_r_fc", shape: [1, 1] - Estimated roll (in degrees).

Output layer names in Caffe* format:

  1. name: "fc_y", shape: [1, 1] - Estimated yaw (in degrees).
  2. name: "fc_p", shape: [1, 1] - Estimated pitch (in degrees).
  3. name: "fc_r", shape: [1, 1] - Estimated roll (in degrees).

Each output contains one float value that represents value in Tait-Bryan angles (yaw, pitсh or roll).

Legal Information

[*] Other names and brands may be claimed as the property of others.