eric-ai-lab / VLMbench

NeurIPS 2022 Paper "VLMbench: A Compositional Benchmark for Vision-and-Language Manipulation"
MIT License
79 stars 8 forks source link

An issue about the obs #13

Open Zhangwenyao1 opened 3 months ago

Zhangwenyao1 commented 3 months ago

Thanks for your great work! Are there any segmentation masks or bbox in the dataset? And if you will consider add these annotations into the dataset or are there interfaces to extract these annotations in this work?

KzZheng commented 3 months ago

Yes, there are segmentations in the dataset, and you can find all observation keys at line 11 of examples/multi_task.py