how to train this network?

[MedVisionKitchen]

Hi, I am a new learner of 3d human pose estimation. This code is really helpful. I want to know how to train my own datasets, especially when the output of the network is more than one guy?

[Daniil-Osokin]

Hi! The training code has not been released. However you can use neighbour repository, it gives 2D multi-person pose estimation, and add 3D coordinates prediction as described in the paper Single-Shot Multi-Person 3D Pose Estimation From Monocular RGB