search

AI-liu / Complex-YOLO

Complex-YOLO: Real-time 3D Object Detection on Point Clouds pytorch Darknet
435 stars 113 forks source link

training on own data: classes #13

Closed yerzhik closed 5 years ago

yerzhik commented 5 years ago

How do I train on own data, I already generated data/labels, but in my case I have 3 classes. So I decided to run and changed the

region_loss = RegionLoss(num_classes=8, num_anchors=5) to region_loss = RegionLoss(num_classes=3, num_anchors=5) and got an error: RuntimeError: shape '[12, 5, 10, 16, 16]' is invalid for input of size 230400

seems like second and third element of the output shape should be equal to 75 if multiplied together. But not sure what should I change to introduce different amount of classes.

yerzhik commented 5 years ago

So I found it out, number of classes is 3 then num_class=4 and at the region_loss.py make the output.view third parameter to be 15 in sum (numclasses + 11).

abhigoku10 commented 5 years ago

@yerzhik where you able to get class predictions properly after training ?? can you pls share the difference you made

yerzhik commented 5 years ago

@abhigoku10 Sorry, what exactly do you mean? Changes I've made to adapt to different number of classes?

abhigoku10 commented 5 years ago

@yerzhik when I train the model it gives negative class prediction probability. Where u getting the same, if no not what changes have you made