jsll / pytorch_6dof-graspnet

MIT License
146 stars 55 forks source link

How do I get the point cloud data from Realsense and convert it to the NPY files needed for the project demo? #7

Open gzchenjiajun opened 3 years ago

gzchenjiajun commented 3 years ago

I am using your warehouse, I tried the demo effect and felt very good. I see that the demo code is an input of test data in NPY format, and the output is the score and grasp location.

Now I want to merge the data from Intel RealSense into a similar output format, and then give it to machine control. How do I do that?

gzchenjiajun commented 3 years ago

Smoothed pc comes from averaging the depth for 10 frames and removing

the pixels with jittery depth between those 10 frames.

How exactly is this step generated? Is there a reference case?

gzchenjiajun commented 3 years ago

Smoothed pc comes from averaging the depth for 10 frames and removing

the pixels with jittery depth between those 10 frames.

How exactly is this step generated? Is there a reference case?

I understand it's 10 frames of point cloud data, right?

jsll commented 3 years ago

Yes you are correct it is 10 frames of point-cloud data.

Now I want to merge the data from Intel RealSense into a similar output format, and then give it to machine control. How do I do that?

What you need to do is to read in the data from Intel RealSense and then convert these to pytorch tensors which is then fed to the network.

gzchenjiajun commented 3 years ago

Yes you are correct it is 10 frames of point-cloud data.

Now I want to merge the data from Intel RealSense into a similar output format, and then give it to machine control. How do I do that?

What you need to do is to read in the data from Intel RealSense and then convert these to pytorch tensors which is then fed to the network.

Thank you very much, I will try it these days! Your Git project is really great! I can't wait to actually use it

gzchenjiajun commented 3 years ago

What does the output value of the network xyz stand for? Is the output XYZ based on the xyz value in the center of the object or on the XYZ value in the bottom left corner of the image? thank you

@jsll

jsll commented 3 years ago

xyz is the 3D coordinates of the point-cloud

gzchenjiajun commented 3 years ago

link:https://pan.baidu.com/s/1Q8Xbk9sW2sZ50Wk_aUz4LA password:coo3

I wrote my own code to gather data from Realsense and infer, So here's the code. But now I have made a mistake, I don't know how to solve it, could you help me? thank you

image

imdoublecats commented 3 years ago

link:https://pan.baidu.com/s/1Q8Xbk9sW2sZ50Wk_aUz4LA password:coo3

I wrote my own code to gather data from Realsense and infer, So here's the code. But now I have made a mistake, I don't know how to solve it, could you help me? thank you

image

Your link is dead

gzchenjiajun commented 3 years ago

@imdoublecats 我换了另一个repo去跑,你可以帮帮我吗?谢谢 ,链接在下面 https://github.com/NVlabs/6dof-graspnet/issues/11#issuecomment-752355589