EZS-Bench_data

# EZS-Bench Data Evaluation data for **EZS-Bench** (Embodied Zero-Shot Benchmark), a benchmark for assessing video generation methods in robotic manipulation scenarios. ## Contents - `video_prompt_question_196_ezs0.jsonl` — 196 evaluation samples, each containing: - `video`: path to the conditioning image - `prompt`: detailed scene description for video generation - `question`: VQA questions for evaluating robot fidelity - `images/` — 196 conditioning images corresponding to each sample ## Usage Use with the [EZS-Bench](https://github.com/amap-cvlab/ABot-PhysWorld/tree/main/EZS-Bench) evaluation toolkit: ```bash python evaluate_ezsbench.py \ --data_file video_prompt_question_196_ezs0.jsonl \ --method_name "YourMethod" \ --method_dir /path/to/generated_videos \ --output_dir ./results ``` ## Citation ```bibtex @article{abot-physworld2026, title={ABot-PhysWorld: Interactive World Foundation Model for Robotic Manipulation with Physics Alignment}, author={Yuzhi Chen, Ronghan Chen, Dongjie Huo, Yandan Yang, Dekang Qi, Haoyun Liu, Tong Lin, Shuang Zeng, Junjin Xiao, Xinyuan Chang, Feng Xiong, Xing Wei, Zhiheng Ma, Mu Xu}, year={2026} } ```