Vegeta2020 / SE-SSD

SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud, CVPR 2021.
Apache License 2.0
811 stars 128 forks source link

The following error occurred when I compiled and installed iou3d #76

Open libingDY opened 2 years ago

libingDY commented 2 years ago

g++: error: /home/libing/newstorage/3Dobjectdetector/SE-SSD/det3d/core/iou3d/build/temp.linux-x86_64-3.7/src/iou3d.o: 没有那个文件或目录 g++: error: /home/libing/newstorage/3Dobjectdetector/SE-SSD/det3d/core/iou3d/build/temp.linux-x86_64-3.7/src/iou3d_kernel.o: 没有那个文件或目录 g++: error: /home/libing/newstorage/3Dobjectdetector/SE-SSD/det3d/core/iou3d/build/temp.linux-x86_64-3.7/src/iou3d_cpu.o: 没有那个文件或目录 error: command 'g++' failed with exit status 1

I can't compile those three files

My environment is: cuda 10.1 pytorch 1.7

I can't compile the following three files

What is the configuration of your environment?

Vegeta2020 commented 2 years ago

@libingDY You may find the env config in https://github.com/poodarchu/Det3D/blob/master/INSTALLATION.md

For my env, the cuda 10.1 and pytorch 1.4.0 is fine, but I have not used so high version of pytorch as 1.7.

libingDY commented 2 years ago

Thank you very much for your reply. I have changed the version of torch, but I have encountered the following problems:

True Traceback (most recent call last): File "train.py", line 18, in from det3d.torchie.apis import ( File "/home/libing/newstorage/3Dobjectdetector/SE-SSD/det3d/torchie/apis/init.py", line 5, in from .train_sessd import batch_processor, build_optimizer, train_detector File "/home/libing/newstorage/3Dobjectdetector/SE-SSD/det3d/torchie/apis/train_sessd.py", line 7, in import apex File "/home/libing/anaconda3/envs/HVPR/lib/python3.7/site-packages/apex/init.py", line 23, in from . import transformer File "/home/libing/anaconda3/envs/HVPR/lib/python3.7/site-packages/apex/transformer/init.py", line 1, in from apex.transformer import amp File "/home/libing/anaconda3/envs/HVPR/lib/python3.7/site-packages/apex/transformer/amp/init.py", line 1, in from apex.transformer.amp.grad_scaler import GradScaler File "/home/libing/anaconda3/envs/HVPR/lib/python3.7/site-packages/apex/transformer/amp/grad_scaler.py", line 8, in class GradScaler(torch.cuda.amp.GradScaler): AttributeError: module 'torch.cuda' has no attribute 'amp'

Vegeta2020 commented 2 years ago

@libingDY It seems an issue of apex, you may also need to reinstall it after updating your pytorch version.

LiuPearl1 commented 2 years ago

@Vegeta2020 My environment is the cuda 10.1 and pytorch 1.4.0, And I meet "AttributeError: module 'torch.cuda' has no attribute 'amp'" problem, should I update my torch version? But high version torch will meet build problem like #22 . So which version is right?

346575116 commented 1 year ago

请问这个问题解决了吗?我也遇到了同样的问题,我的环境是cuda10.0和pytorch1.1.0,尝试升级pytorch1.7也会带来其他问题。

qq123aa456 commented 1 year ago

@LiuPearl1 have you solved it?I meet the same question which puzzled me long time.