Related to this topic, Facebook AI research recently published an interesting paper on human pose estimation in CNNs:
https://arxiv.org/abs/1802.00434
ntm
It's also really exciting that OpenPose and DensePose (the work mentioned above from FAIR) have enabled applications like video-to-video synthesis! This presents some interesting potentials to develop photorealistic imagined outputs from a variety of source videos.
We could try (install) it by this.
Related to this topic, Facebook AI research recently published an interesting paper on human pose estimation in CNNs: https://arxiv.org/abs/1802.00434
It's also really exciting that OpenPose and DensePose (the work mentioned above from FAIR) have enabled applications like video-to-video synthesis! This presents some interesting potentials to develop photorealistic imagined outputs from a variety of source videos.