Motion Representations for Articulated Animation

22 Apr 2021 • Aliaksandr Siarohin • Oliver J. Woodford • Jian Ren • Menglei Chai • Sergey Tulyakov

We propose novel motion representations for animating articulated objects consisting of distinct parts. In a completely unsupervised manner, our method identifies object parts, tracks them in a driving video, and infers their motions by considering their principal axes... (read more)

PDF Abstract PDF Abstract

Datasets


Introduced in the Paper:

TED-talks

Mentioned in the Paper:

VoxCeleb1 MGif Tai-Chi-HD
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Video Reconstruction MGif Siarohin et al. L1 0.0206 # 1
Video Reconstruction MGif FOMM L1 0.0223 # 2
Video Reconstruction Tai-Chi-HD (256) Siarohin et al. L1 0.047 # 1
AED 0.152 # 1
AKD 5.58 # 1
MKR 0.027 # 1
Video Reconstruction Tai-Chi-HD (256) FOMM L1 0.056 # 2
AED 0.172 # 2
AKD 6.53 # 2
MKR 0.033 # 2
Video Reconstruction Tai-Chi-HD (512) Siarohin et al. L1 0.064 # 1
AKD 13.86 # 1
MKR 0.043 # 1
AED 0.172 # 1
Video Reconstruction Tai-Chi-HD (512) FOMM L1 0.075 # 2
AKD 17.12 # 2
MKR 0.066 # 2
AED 0.203 # 2
Video Reconstruction TED-talks FOMM L1 0.033 # 2
AKD 7.07 # 1
MKR 0.014 # 1
AED 0.163 # 1
Video Reconstruction TED-talks Siarohin et al. L1 0.026 # 1
AKD 3.75 # 2
MKR 0.007 # 2
AED 0.114 # 2
Video Reconstruction VoxCeleb FOMM L1 0.041 # 2
AKD 1.27 # 1
AED 0.134 # 2
Video Reconstruction VoxCeleb Siarohin et al. L1 0.040 # 1
AKD 1.28 # 2
AED 0.133 # 1

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet