Face recognition (identification) after detection

[schelian]

Hi all, has anyone done this? I am using the OpenFace and its ROS wrapper. This wants a 2D image but I just want to pass it the ROI / bounding box where the face was first detected (plus a little area around it).

The messages output from this package are the 3D coordinate but the visualization part of it has an ROI / bounding box. Should I change the message from this package to also output the ROI or is there an easier way?

[DLu]

I don't believe anyone is actively working on this, at least as part of this stack.