Work

Improving Human Pose-Conditioned Generation: Fine-Tuning ControlNet Models with Reinforcement Learning

FWCI: 1.57
Cites: 38
Cited by: 2
Related to: 10