MoLA: Motion Generation and Editing with Latent Diffusion Enhanced by Adversarial Training.
Kengo UchidaTakashi ShibuyaYuhta TakidaNaoki MurataShusuke TakahashiYuki MitsufujiPublished in: CoRR (2024)
Keyphrases
- image sequences
- motion model
- motion analysis
- motion estimation
- space time
- human motion
- optical flow
- training data
- training algorithm
- training process
- motion patterns
- anisotropic diffusion
- motion detection
- motion segmentation
- motion field
- training set
- motion parameters
- spatial and temporal
- training examples
- online learning
- feature space
- motion capture
- surgical training