Hi, if I know the instance id of target object, and want to obtain the grasp pose of the target object, is this method feasible? I make modifications in the code of icg_benchmark/grasping/planners/icgnet.py
ori, ctc, score, width, instance = output.scene_grasp_poses
# numpy_to_ply(ctc.cpu().numpy(), '/usr/stud/dira/GraspInClutter/icg_benchmark/demo/demo_ctc.ply')
# mask = score > self.confidence_th
mask = (score > self.confidence_th) & (instance == targ_instance_id)
if not mask.any():
print(
colored(
"no candidates with high score" + f"{str(score.max()) if len(score) > 0 else 'No pts'}",
"yellow",
)
)
return None
Hi, if I know the instance id of target object, and want to obtain the grasp pose of the target object, is this method feasible? I make modifications in the code of icg_benchmark/grasping/planners/icgnet.py