MBench Leaderboard

Welcome to the MBench Leaderboard!

On this leaderboard we share the evaluation results of MBench proposed in our paper:

The Quest for Generalizable Motion Generation: Data, Model, and Evaluation

Code arXiv MBench-Leaderboard ViMoGen-228K

This leaderboard evaluates Motion Generation models on comprehensive metrics covering:

  • VLM-based metrics: Motion Generalizability, Motion-Condition Consistency
  • Physics-based metrics: Jitter Degree, Dynamic Degree, Foot Floating, Foot Sliding, Body Penetration
  • Distribution-based metrics: Pose Quality
{
  • "headers": [
    • "Method",
    • "Motion Generalizability ↑",
    • "Motion-Condition Consistency ↑",
    • "Jitter Degree ↓",
    • "Dynamic Degree ↑",
    • "Foot Floating ↓",
    • "Foot Sliding ↓",
    • "Ground Penetration ↓",
    • "Body Penetration ↓",
    • "Pose Quality ↓"
    ],
  • "data": [
    • [
      • "<a target="_blank" href="https://github.com/oneScotch/ViMoGen" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">ViMoGen</a>",
      • 0.68,
      • 0.53,
      • 0.0108,
      • 0.0251,
      • 0.204,
      • 0.0064,
      • 0.076,
      • 1.78,
      • 2.38
      ],
    • [
      • "<a target="_blank" href="https://github.com/oneScotch/ViMoGen" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">ViMoGen-light</a>",
      • 0.55,
      • 0.47,
      • 0.0129,
      • 0.0294,
      • 0.155,
      • 0.0051,
      • 0.059,
      • 1.43,
      • 2.1
      ],
    • [
      • "<a target="_blank" href="https://github.com/Dai-Wenxun/MotionLCM" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">MotionLCM</a>",
      • 0.55,
      • 0.48,
      • 0.0218,
      • 0.0439,
      • 0.193,
      • 0.0202,
      • 0.1507,
      • 1.73,
      • 2.4
      ],
    • [
      • "<a target="_blank" href="https://github.com/GuyTevet/motion-diffusion-model" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">MDM</a>",
      • 0.51,
      • 0.42,
      • 0.0136,
      • 0.0376,
      • 0.156,
      • 0.0136,
      • 0.134,
      • 1.68,
      • 2.67
      ],
    • [
      • "<a target="_blank" href="https://github.com/cure-lab/MotionCraft" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">MotionCraft</a>",
      • 0.45,
      • 0.42,
      • 0.0132,
      • 0.042,
      • 0.402,
      • 0.009,
      • 0.137,
      • 1.15,
      • 2.12
      ],
    • [
      • "<a target="_blank" href="https://github.com/EricGuo5513/momask-codes" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">MoMask</a>",
      • 0.44,
      • 0.38,
      • 0.0147,
      • 0.0396,
      • 0.178,
      • 0.0147,
      • 0.168,
      • 1.48,
      • 2.67
      ],
    • [
      • "<a target="_blank" href="https://github.com/mingyuan-zhang/FineMoGen" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">FineMoGen</a>",
      • 0.42,
      • 0.37,
      • 0.0118,
      • 0.0386,
      • 0.281,
      • 0.0091,
      • 0.141,
      • 1.18,
      • 2.28
      ],
    • [
      • "<a target="_blank" href="https://github.com/mingyuan-zhang/MotionDiffuse" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">MotionDiffuse</a>",
      • 0.42,
      • 0.44,
      • 0.0111,
      • 0.0289,
      • 0.126,
      • 0.0063,
      • 0.076,
      • 1.35,
      • 2.21
      ],
    • [
      • "<a target="_blank" href="https://github.com/Mael-zys/T2M-GPT" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">T2M-GPT</a>",
      • 0.38,
      • 0.39,
      • 0.0156,
      • 0.0349,
      • 0.209,
      • 0.0156,
      • 0.1248,
      • 1.33,
      • 2.43
      ]
    ],
  • "metadata": null
}