-
Hello, my dear friend, could I know how to test one module's runtime or inferring time, such as PV RCNN?
-
I'm getting this error while trying to use the prebuild
`C:\Users\91828\Desktop\OpenPCDet>python setup.py develop
running develop
C:\Users\91828\AppData\Local\Programs\Python\Python38\lib\site-…
-
How should I solve this issue
```
Traceback (most recent call last):
File "test.py", line 9, in
from pcdet.datasets.my_custom_dataset import MyCustomDataset
File "/Dev/xelawk_dev/OpenPCD…
-
Thanks for your wonderful work~
I am puzzled on the training strategy. The PointRCNN paper says that
> The two stage sub-networks of PointRCNN are trained separately.
, and the [official cod…
-
https://github.com/open-mmlab/OpenPCDet/blob/e4d2b75bc0e48399177a1823ad6f4ff8179cc835/pcdet/models/dense_heads/anchor_head_template.py#L43
In convolution downsampling with stride 2, if original fea…
-
HI,author!
The model learning rate is defined as follows:
"
lr_scheduler = _create_learning_rate_scheduler(
optimizer, cfg.lr_config, total_steps)
def step(self, step):
lrs, …
-
how to understand the four evaluation indicators as follows:
①Car AP@0.70, 0.70, 0.70
②Car AP_R40@0.70, 0.70, 0.70
③Car AP@0.70, 0.50, 0.50
④Car AP_R40@0.70, 0.50, 0.50
the first Car AP@0.70, 0.…
-
Hi I found the `points_in_boxes` functions returns an index tensor indicating the index of box where each point is in, so this function only works when the boxes have no overlap? If the boxes may have…
-
https://github.com/open-mmlab/OpenPCDet/blob/0627984bb8bd9b360de2dee77ca2103472d64e65/pcdet/datasets/pandaset/pandaset_dataset.py#L304
-
I used to use tf-2.4.0 and the corresponding waymo-open-dataset tools. However, I found the trained model get very low mAP (10% on Vehicle). I notice that it is recommended to modify the version in (h…