README.md

metadata

license: cc-by-nc-4.0
pretty_name: RoboJudge-Bench
task_categories:
  - video-classification
  - visual-question-answering
language:
  - en
tags:
  - robojudge
  - embodied-ai
  - video-evaluation
  - benchmark

RoboJudge-Bench

RoboJudge-Bench contains 800 human-annotated embodied videos from 8 source corpora and 10 video generators. Every record includes the task instruction, locally hosted video and initial-frame URLs, PA/IA ratings, and six diagnostic sub-scores.

Files

• data/test.jsonl: benchmark metadata, instructions, links, and labels.
• data/<source>/generated_data/.../*.mp4: all 800 benchmark videos.
• data/<source>/generated_data/.../prompt/: initial frames and instructions.
• metadata/summary.json: release counts and source revision.

Scores

• Physical Adherence (PA): agent consistency, scene consistency, and interaction realism.
• Instruction Alignment (IA): agent match, object correctness, and goal completion.

Main scores are integers from 1 to 5. Diagnostic sub-scores are integers from 0 to 2.

Project page: https://siyuanmacs.github.io/RoboJudge/