3D目标检测方法TED复现：Transformation-Equivariant 3D Object Detection for Autonomous Driving

最新推荐文章于 2025-05-18 20:17:36 发布

xyhaaab

最新推荐文章于 2025-05-18 20:17:36 发布

阅读量205

点赞数 1

文章标签：目标检测人工智能计算机视觉

本文链接：https://blog.csdn.net/xyhaaab/article/details/147197266

版权

论文链接：[2211.11962] Transformation-Equivariant 3D Object Detection for Autonomous Driving

项目地址：hailanyi/TED: Transformation-Equivariant 3D Object Detection for Autonomous Driving

写在前面：原项目在V100和3090上测试，但是推荐的cuda11.1不支持4090

1.复现环境：

Ubuntu 20.04
Python 3.9.21
PyTorch 2.6.0
Spconv 2.3.8
NVIDIA CUDA 11.8
4090 GPU

2.环境配置：

conda create -n spconv2 python=3.9

conda activate spconv2

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

pip install numpy==1.19.5 protobuf==3.19.4 scikit-image==0.19.2 nuscenes-devkit==1.0.5 numba==0.56.4 open3d==0.16.0 scipy pyyaml easydict fire tqdm shapely matplotlib opencv-python addict pyquaternion awscli pandas future pybind11 tensorboardX tensorboard Cython prefetch-generator spconv-cu118 

pip install tensorflow==2.5.0 --index-url https://mirrors.aliyun.com/pypi/simple/

pip install keras-nightly==2.5.0.dev --index-url https://mirrors.aliyun.com/pypi/simple/

pip install waymo-open-dataset-tf-2-5-0

接下来就和github上的README一样啦！