This is a multi-person 2D pose estimation network based on the EfficientHRNet approach (that follows the Associative Embedding framework). For every person in an image, the network detects a human pose: a body skeleton consisting of keypoints and connections between them. The pose may contain up to 17 keypoints: ears, eyes, nose, shoulders, elbows, wrists, hips, knees, and ankles.
|Average Precision (AP)||54.3%|
Average Precision metric described in COCO Keypoint Evaluation site.
1, 3, 448, 448 in the
B, C, H, W format, where:
B- batch size
C- number of channels
H- image height
W- image width
Expected color order is
The net outputs are two blobs:
1, 17, 224, 224containing location heatmaps for keypoints of all types. Locations that are filtered out by non-maximum suppression algorithm have negated values assigned to them.
1, 17, 224, 224, 1containing associative embedding values, which are used for grouping individual keypoints into poses.
[*] Other names and brands may be claimed as the property of others.