search

rohitgirdhar / CATER

CATER: A diagnostic dataset for Compositional Actions and TEmporal Reasoning
https://rohitgirdhar.github.io/CATER/
Apache License 2.0
103 stars 19 forks source link

How to get the GT masks of objects? #15

Closed sjtuytc closed 4 years ago

sjtuytc commented 4 years ago

Hi, I want to run some segmentation codes (e.g. Mask RCNN) so I need the gt masks of objects in the scene. CLEVRER has provided them but CATER does not. I want to know do you plan to release the masks of objects? If not, how can we get them?

rohitgirdhar commented 4 years ago

In the pre-generated dataset, unfortunately there's no easy way to get segmentation maps (more discussion in https://github.com/rohitgirdhar/CATER/issues/7). It should be possible to generate it when generating the data though, for eg, by rendering additional videos with 1 object at a time without lighting in a uniform color, and thresholding (similar to how CLEVR checks for visibility).