Giters

vimalabs / VIMABench

Official Task Suite Implementation of ICML'23 Paper "VIMA: General Robot Manipulation with Multimodal Prompts"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Question about getCameraImage function

XihangYU630 opened this issue · comments

commented

Why in the function p.getCameraImage, it sets flags=p.ER_SEGMENTATION_MASK_OBJECT_AND_LINKINDEX,
renderer=p.ER_BULLET_HARDWARE_OPENGL?

_, _, color, depth, segm = p.getCameraImage(
width=image_size[1],
height=image_size[0],
viewMatrix=viewm,
projectionMatrix=projm,
flags=p.ER_SEGMENTATION_MASK_OBJECT_AND_LINKINDEX,
renderer=p.ER_BULLET_HARDWARE_OPENGL,
physicsClientId=self.client_id,
)

commented

Thanks for the question. We set ER_SEGMENTATION_MASK_OBJECT_AND_LINKINDEX because we also want to provide ground-truth object segmentation. For ER_BULLET_HARDWARE_OPENGL because we use headless rendering. We'd recommend PyBullet doc for more details.