Face detector for driver monitoring and similar scenarios. The network features a default MobileNet backbone that includes depth-wise convolutions to reduce the amount of computation for the 3x3 convolution block.
Metric | Value |
---|---|
AP (head height >10px) | 37.4% |
AP (head height >32px) | 84.8% |
AP (head height >64px) | 93.1% |
AP (head height >100px) | 94.1% |
Min head size | 90x90 pixels on 1080p |
GFlops | 2.835 |
MParams | 1.053 |
Source framework | Caffe* |
Average Precision (AP) is defined as an area under the precision/recall curve. Numbers are on Wider Face validation subset.
Link to performance table
-
name: "input" , shape: [1x3x384x672] - An input image in the format [BxCxHxW], where:
- B - batch size
- C - number of channels
- H - image height
- W - image width
Expected color order is BGR.
- The net outputs blob with shape: [1, 1, N, 7], where N is the number of detected
bounding boxes. For each detection, the description has the format:
[
image_id
,label
,conf
,x_min
,y_min
,x_max
,y_max
]image_id
- ID of the image in the batchlabel
- predicted class IDconf
- confidence for the predicted class- (
x_min
,y_min
) - coordinates of the top left bounding box corner - (
x_max
,y_max
) - coordinates of the bottom right bounding box corner.
[*] Other names and brands may be claimed as the property of others.