Bichen Wu

We provide a converted KITTI data that you can directly download. The 5 dimensions include: x, y, z, intensity, mask, where the mask indicates if this is a missing pixel...

Thanks for your question. I'll look into it. @cksl

Similar questions have been asked. Please take a look at closed issues and see previous discussions.

Thanks for your question. That's a bug due to a recent update. Now it should be fixed. Could you please pull the update and try again?

One other point I can think of is you need to change the number of anchors according to you input resolution. And welcome to share your experience or implementation afterward.

Thanks for your question and discovery about the JIT compilation. > self._add_forward_graph() self._add_interpretation_graph() self._add_loss_graph() self._add_train_graph() self._add_viz_graph() Not all of them are related to inference. The minimal operations you need is...

@aswwqhome1993 If you could provide more details, it would be helpful. - Did you re-train the model? If so, how many iterations did you train? Did you make any modifications...

Seems that the model is somehow broken. I will train a new model and update it in next few days. Thanks.

@wm901115nwpu That is expected since SqueezeDet is loading parameter initialization from conv1 - fire9 from a pre-trained SqueezeNet model, while fire10, fire11, fire12 are randomly initialized.

Thanks for letting me know. Seems that it's broken by a recent update. I'll fix that later.