Lidar_AI_Solution

Lidar_AI_Solution copied to clipboard

which model should I use?

5

Thanks for your great work, but I really need some advices! I want to deploy a lidar 3d detector,which model should I choose? BEVFusion or CenterPoint? Now, I only get...

mx2013713828

How should I export the data of camera_bevfeat of type const nvtype::half*, because I want to visualize the BEV results of Camera. Thank you very much.

2

Here is the code. I hope to view the camera_bevfeat data through csv file, it seems that the output dimension is {1x80x180x180}[float16]. const nvtype::half* forward_only(const void* camera_images, const nvtype::half* lidar_points,...

KaysenC

why use "val = val.permute(4, 0, 1, 2, 3)" when export scn to onnx in centerpoint net

1

hello, i wanna know why use val = val.permute(4, 0, 1, 2, 3) when export scn to onnx in centerpoint, all pth files are generated in sp2.X thank you!

pangweisong

how to train and use myself data

1

I want to use myself dataset, and multi camera(set the number with config)，and it maybe need retrain,how to do,thank you.

xlmxbw

CUDA Runtime error cudaEventCreate(&begin_) # the provided PTX was compiled with an unsupported toolchain., code = cudaErrorUnsupportedPtxVersion [ 222 ] in file /home/ubuntu/tensorRT/chapter8-learn-from-oss/Lidar_AI_Solution/CUDA-BEVFusion/src/common/timer.hpp:34 tool/run.sh: line 44: 255102 Aborted (core dumped)...

jiafuzeng

Hi! First of all, thanks for your great work. I have successfully tested BEVFusion with the original Nuscenes dataset and trained the network with a CARLA dataset using swint_v0p075. I...

jorgemn9

why not use TensorRT for LidarBackbone model in CUDA-BEVFusion?

1

Hello, In CUDA-BEVFusion, LiDAR Backbone uses ONNX format files (lidar.backbone.xyz.onnx). Is there any reason why the TensorRT transformation is not performed as in other models (e.g. camera.backbone.onnx)?

fsi-kawatawa

add node cause result wrong!

1

identity = x.features out = self.conv1(x) out = out.replace_feature(self.relu1(out.features)) out = self.conv2(out) out = out.replace_feature(out.features+identity) out = out.replace_feature(self.relu2(out.features)) if without "out = out.replace_feature(out.features+identity) " ,the c++ and pytroch result is...

NotOnlyAI

libspconv source code

1

Would you please open source the spconv code? It would be very helpful for the community.

jongsik-moon

CUDA-BEVFusion: difference in the result of onnx and pytorch model

5

Hi. I have converted BEVFusion model from torch to onnx but it fails np.testing.allclose ``` Mismatched elements: 17068436 / 100663296 (17%) Max absolute difference: 0.00343999 Max relative difference: 17035.555 x:...

Jayden9912