Lidar_AI_Solution

Lidar_AI_Solution copied to clipboard

an illegal memory access was encountered

3

when I try to inference bevfusion model on more than 800000 points, an error occurred and core dumped: CUDA Runtime error cudaStreamSynchronize(stream) # an illegal memory access was encountered, code...

erzhu222

Dear author: I use your CUDA_BEVFusion to perform inference on my device, but the output results obtained are all NAN. Do you know why this is? I even changed devices...

chenzb0712

Support for H100 and sm_90

2

Hi, Does this project work on the H100? It seems that libspconv.so does not support sm_90. Is it possible to get a binary that supports it? ``` $ cuobjdump --list-elf...

maueki

Hi, Thanks for the great work. When I tried to deploy libspconv in a real-time autonomous driving system, I had to constantly copy the sparse network's inference output into a...

ThomasMouse

Does the point cloud range need to be symmetrical from zero?

1

Hi, When I use `point_cloud_range=[-54, -100, -5 54, 100, 3]`, the prediction results are quite reasonable. - `point_cloud_range=[-54, -100, -5 54, 100, 3]` - `voxel_size=[0.075, 0.1389, 0.2]` However, in the...

san9569

Spconv Engine Implementation

5

Hi @hopef ! I found a huge gap of inference latency on orin of BEVFusion's lidar-scn backbone, between using **the engine returned by spconv::load_engine_from_onnx(30ms)** and using **the original pytorch implementation(100ms)**....

barrydoooit

when I inference the same Bin file, The number of voxels not equal! while I use Centerpoint , I got 87081 voxels after voxel_generator, and 111576 voxels in CUDA-Centerpoint after...

mx2013713828

Hi, @hopef This program requires both point cloud and images. Is it possible to have only point cloud or images, so that the program's robustness may be better?

MarioUncle-Wang

I want to use custom data, two ways. First one, transfer kitti type data to nuscenes, I dont find correct way, failed. Second one, train centerpoint with kitti, I get...

ruishanyin

i am working on CUDA_Bevfusion and i am getting error regarding cuda, cuDNN and Tensorrt, So please can anyone provide dockerfile ??? Thanks in Advance...

Ankitzanzmera