This is a 2D facial landmarks detection network based on the HRNet approach. For face in an image, the network detects landmarks (look at image below). The landmarks contain 98 keypoints.
Network is trained and validated on the custom dataset based on WiderFace and VGG2 subsets.
| Metric | Value |
|---|---|
| NME | 0.1323 |
| GFlops | 0.6 |
| MParams | 9.66 |
| Source framework | PyTorch* |
Name: input.1, shape: 1, 3, 64, 64. An input image in the B, C, H, W format, where:
B- batch sizeC- number of channelsH- image heightW- image width Expected color order isBGR.
The net outputs a blob 3851 with the shape: 1, 98, 16, 16, containing location heatmaps for 98 keypoints. Locations that are filtered out by non-maximum suppression algorithm have negated values assigned to them.
The model can be used in the following demos provided by the Open Model Zoo to show its capabilities:
[*] Other names and brands may be claimed as the property of others.