How do I get keypoints extraction? about dataset_synthesizer

[darwinharianto]

@yangninghua So instead of having one object, I created multiple objects. One object would be the main object, the others would be the 'keypoints'. Parenting other objects to main Objects would make them move together.
The resulting annotation would be Main object + other objects, and I just have to create a script that takes the other objects location as the keypoints. Hacky solution, but it works...

[kaixin-bai]

Since you already have the position of neck, hands, legs of the doll in object coordinate system, and you can also get the translation and rotation of the doll in camera coordinate system with NDDS, maybe you can just convert the coordinate system from object to camera, then you can get the position of keypoints in camera coordinate system.

[HaiyiMei]

@darwinharianto @kaixin-bai Hey guys, how did you solve this problem? I'm facing the same issues now. Thanks!

[darwinharianto]

@HaiyiMei Sorry for the late reply, I ended up making a cube and parented them inside object. Later on I used another script to convert them to COCO format.

[yangninghua]

@HaiyiMei Sorry for the late reply, I ended up making a cube and parented them inside object. Later on I used another script to convert them to COCO format.

Excuse me, what does this mean? I don't quite understand it