whats then meaning of each number of rect grasp label

[cww97]

>>> f = '/home/weiwen/mnt0/mail2020/graspDL/datasets/graspNet-1b/scenes/scene_0000/kinect/rect/0000.npy'
>>> rect = np.load(f)
>>> rect.shape                                                                                           
(7645, 7) 

[Fang-Haoshu]

Hi, should be (score, width, height, depth, rotation, translation, object_id), where object id is set to -1 by default(not used)

[gachiemchiep]

Hello @Fang-Haoshu I understand the meaning of score, rotation, translation, object_id. Would you mind explain the meaning of width, height, depth. Is the width equal to "6", height equal to "7" and depth equal to "4" as in this picture.

[Fang-Haoshu]

Hi, it is like this one

[cww97]

I have successfully get rect grasps from here, does these 7-d rect grasp the same as what you explained above?

further more, I wanna evaluate these grasps, can I use convert from here, and then use evaluation from here?

[gachiemchiep]

@Fang-Haoshu Sorry, I still don't understand those values. For example, if we detected a grasp pose like this image : https://github.com/atenpas/gpd/blob/master/readme/hand_frame.png For now, I understand the meaning of each variable like this:

1. score: the confidence of detected grasp
2. rotation: rotation matrix of detected grasp
3. translation: position (3d coordinates) where grasp is detected
4. object_id
5. width, height, depth: I guess that is the size of cube that contains gripper