search

NVlabs / contact_graspnet

Efficient 6-DoF Grasp Generation in Cluttered Scenes
Other
324 stars 111 forks source link

Scores model output #27

Closed danialdunson closed 1 year ago

danialdunson commented 2 years ago

I have noticed that the model produces scores when segmentation mask is set to None. Do these scores have any significance and if so what is that?

P.S. I added your model as a servicecall in a ROSBRIDGE, and, currently, I am naively selecting the top 5 poses with the highest score values without a segmentation mask. Also, i added the the appropriate transforms to get it working on the interbotix wx250s. its surprisingly good even before training and just filtering out invalid gripper widths. I am a senior in undergrad btw, so my journey has been similar to: https://github.com/NVlabs/contact_graspnet/issues/8

MartinSmeyer commented 2 years ago

Hey @danialdunson,

sorry, missed your issue over the summer somehow. If you do not provide a segmentation mask the network will not focus on a particular object but predict grasps for the entire scene. It's even better though if it can focus as described here.