search

yzrobot / online_learning

[ROS package] Online Learning for Human Detection in 3D Point Clouds
GNU General Public License v3.0
156 stars 70 forks source link

Train my own SVM model #22

Closed minghaohsu410168 closed 3 years ago

minghaohsu410168 commented 3 years ago

excuse me,

I have played my own bag file collected from VLP16, but the result is poor, may I ask how to train my own svm model by my self? I am a newer with ml.

thank you

yzrobot commented 3 years ago

Hi, you could check this: https://github.com/yzrobot/cloud_annotation_tool/tree/devel, hope it helps. Best

minghaohsu410168 commented 3 years ago

Thank you for reply, but I can't use the cloud_annotaation_tool, does it not support ubuntu18.04?

and do you have the detail code or step about training the svm? I am practicing model training.

akamob commented 3 years ago

@minghaohsu410168 Hi, I meet the same problem and I have tried the point cloud annotation tool but I haven’t got a clue about how to run it to train the svm model.

May I have your suggestions to fix this problem? Any help is much appreciated:)

minghaohsu410168 commented 3 years ago

@akamob You can use the devel branch provided by the author. If the repo cannot be built successfully, try to build in Ubuntu 14. Hope it helps.

akamob commented 3 years ago

@minghaohsu410168 Thank you for your reply,

Now I can use the devel branch repository, but I don't know how to add the negative examples. I guess I should modify viewer.cpp, but after reading it, I haven’t got a clue.

Would it be too much trouble for you to help me understand the code? Any suggestions is much appreciated:)

minghaohsu410168 commented 3 years ago

You can modify the CloudViewer::extractButtonClicked() function in viewer.cpp. Adding the writeData() method with -1 label when object class are car, wheel, etc.