OpenPCDet

OpenPCDet copied to clipboard

Hi, first of all, thank you for great work!!! which is very helpful. 1. for multi-modal DL models I can see transfusion and bev fusion under Waymo dataset config. Is...

MingiJeong

pointpillar_newaugs.yaml can not use.

1

i want to ues DATA_AUGMENTOR but this yaml cant use.how can i use DATA_AUGMENTOR?

fffyyy111

training with pvrcnn losses not decreasing as seen in the tensor-board !!!!

1

training with pvrcnn loss not decreasing as seen in the tensor-board (attached) please kindly guide me, whether its correct or not?  as i trained with SECONDIOU, here i can...

mdfaheem786

questions about GETTING_STARTED.md

2

I would like to start using OpenPCDet with the Waymo dataset. Could you please provide more information about the data format that is needed. I have successfully ran the library...

NayaBaslan

when i run tools/demo.py , the bug has triggered: free(): double free detected in tcache 2. I don't know how to solve this bug, please help me

zhangguotai

It is possible to do either or both of the following with the current code to improve predictions: - 1) Test-time augmentation - 2) Model ensembling (like NMS between model...

AlexandreDiPiazza

Abnormal demo.py result using nuscenses dataset pre-trained models

1

This issue is related to [issue 1341](https://github.com/open-mmlab/OpenPCDet/issues/1341), but here we track a different problem. After modify the demo.py a bit in order to run the nuscenses dataset pre-trained model, I...

yaobaishen

How to get validation loss and display it in Tensorboard

8

 The current project contains only training loss and learning rate curves, how can I modify def train_one_epoch() to compute the validation loss during the training session?

Zixiu99

Thank you for the excellent framework for 3D object detection. I’m curious about the architecture of the 2D backbone, especially the BaseBEVBackbone. I would like to understand it more deeply....

nghiabka

Hello, I would like to add MPPNet support for NuScenes. Given that both datasets seem to have similar availability of sequential LIDAR scans, this seems like it would be fairly...

robotville