【visiontransformer】LETR论文解读及代码实战（二）

LETR: Line Segment Detection Using Transformers without Edges

基于vision-transformer/DETR 提取wireframe的网络框架&＃xff0c;截止日前实现了sota性能。

论文&＃xff1a;https://arxiv.org/abs/2101.01909

代码&＃xff1a;https://github.com/mlpc-ucsd/LETR

项目实战&＃xff1a;

&＃xff08;1&＃xff09;构建环境&＃xff1a;

git clone https://github.com/mlpc-ucsd/LETR.git

mkdir -p data
mkdir -p evaluation/data
mkdir -p expconda create -n letr
conda activate letr
conda install -c pytorch pytorch torchvision
conda install cython scipy
pip install -U &＃39;git&＃43;https://github.com/cocodataset/cocoapi.git#subdirectory&＃61;PythonAPI&＃39;
pip install docopt

&＃xff08;2&＃xff09;准备数据&＃xff0c;wireframe raw 数据集和YorkUrban 数据集。

&＃xff08;3&＃xff09;模型训练

# Usage: bash script/*/*.sh [exp name]
bash script/train/a0_train_stage1_res50.sh res50_stage1 # LETR-R50
bash script/train/a1_train_stage1_res101.sh res101_stage1 # LETR-R101

调试过程及问题解决

 &＃xff08;1&＃xff09;ImportError: cannot import name &＃39;_new_empty_tensor&＃39;

Traceback (most recent call last):File "src/main.py", line 13, in import datasetsFile "/home/wsx/0A_DATA/LETR/src/datasets/__init__.py", line 5, in from .coco import build as build_cocoFile "/home/wsx/0A_DATA/LETR/src/datasets/coco.py", line 10, in import datasets.transforms as TFile "/home/wsx/0A_DATA/LETR/src/datasets/transforms.py", line 18, in from util.misc import interpolateFile "/home/wsx/0A_DATA/LETR/src/util/misc.py", line 22, in from torchvision.ops import _new_empty_tensor
ImportError: cannot import name &＃39;_new_empty_tensor&＃39;


solution&＃xff1a;

定位报错文件中File "/home/wsx/0A_DATA/LETR/src/util/misc.py", line 22, in &＃xff0c;将报错文件misc.py中的关于“_new_empty_tensor”的内容注释掉&＃xff0c;如下&＃xff1a;

# if float(torchvision.__version__[:3]) <0.7:# from torchvision.ops import _new_empty_tensor# from torchvision.ops.misc import _output_size

&＃xff08;2&＃xff09;再运行&＃xff1a;ImportError: cannot import name &＃39;_LinearWithBias&＃39;

Traceback (most recent call last):File "src/main.py", line 17, in from models import build_modelFile "/home/wsx/0A_DATA/LETR/src/models/__init__.py", line 2, in from .letr import buildFile "/home/wsx/0A_DATA/LETR/src/models/letr.py", line 15, in from .transformer import build_transformerFile "/home/wsx/0A_DATA/LETR/src/models/transformer.py", line 16, in from .multi_head_attention import MultiheadAttentionFile "/home/wsx/0A_DATA/LETR/src/models/multi_head_attention.py", line 11, in from torch.nn.modules.linear import _LinearWithBias
ImportError: cannot import name &＃39;_LinearWithBias&＃39;


solution&＃xff1a;

定位出错的文件&＃xff0c;  File "/home/wsx/0A_DATA/LETR/src/models/multi_head_attention.py", line 11, in &＃xff0c;将 line 11 的 # from torch.nn.modules.linear import _LinearWithBias 注释掉&＃xff0c;
改为 from torch.nn.modules.linear import Linear&＃xff0c;如下

# from torch.nn.modules.linear import _LinearWithBias
from torch.nn.modules.linear import Linear

同时将引用部分 multi_head_attention.py 的440行 &＃xff0c;

self.out_proj &＃61; _LinearWithBias(embed_dim, embed_dim)改为&＃xff1a;self.out_proj &＃61; Linear(embed_dim, embed_dim)

&＃xff08;3&＃xff09;再运行&＃xff0c;RuntimeError: The NVIDIA driver on your system is too old (found version 10010).

Traceback (most recent call last):File "src/main.py", line 214, in main(args)File "src/main.py", line 21, in mainutils.init_distributed_mode(args)File "/home/wsx/0A_DATA/LETR/src/util/misc.py", line 421, in init_distributed_modetorch.cuda.set_device(args.gpu)File "/home/wsx/anaconda3/envs/letr/lib/python3.6/site-packages/torch/cuda/__init__.py", line 264, in set_devicetorch._C._cuda_setDevice(device)File "/home/wsx/anaconda3/envs/letr/lib/python3.6/site-packages/torch/cuda/__init__.py", line 172, in _lazy_inittorch._C._cuda_init()
RuntimeError: The NVIDIA driver on your system is too old (found version 10010). Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install a PyTorch version that has been compiled with your version of the CUDA driver.


solution&＃xff1a;

查看pytorch 版本

(letr) wsx&＃64;hello:~/0A_DATA/LETR$ python
Python 3.6.4 |Anaconda, Inc.| (default, Mar 13 2018, 01:15:57)
[GCC 7.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> print(torch.__version__)
1.9.0&＃43;cu102


查看CUDA版本

nvidia-smi

 pytorch 是cuda10.2的&＃xff0c;机子的cuda是10.1的&＃xff0c;故卸载pytorch &＃xff0c;重装pytorch&＃43;cuda,

下面pip安装可省略&＃xff0c;直接用conda  

pip3 install torch&＃61;&＃61;1.8.1&＃43;cu101 torchvision&＃61;&＃61;0.9.1&＃43;cu101 -f https://download.py torch.org/whl/cu101/torch_stable.html


(letr) wsx&＃64;hello:~/0A_DATA/LETR$ python -m pip install torch&＃61;&＃61;1.8.1&＃43;cu101 torchvision&＃61;&＃61;0.9.1&＃43;cu101
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
ERROR: Could not find a version that satisfies the requirement torch&＃61;&＃61;1.8.1&＃43;cu101 (from versions: 1.0.0, 1.0.1, 1.0.1.post2, 1.1.0, 1.2.0, 1.3.0, 1.3.1, 1.4.0, 1.5.0, 1.5.1, 1.6.0, 1.7.0, 1.7.1, 1.8.0, 1.8.1, 1.9.0, 1.9.1, 1.10.0, 1.10.1, 1.10.2)
ERROR: No matching distribution found for torch&＃61;&＃61;1.8.1&＃43;cu101


 conda 安装

conda install pytorch&＃61;&＃61;1.8.0 torchvision&＃61;&＃61;0.9.0 torchaudio&＃61;&＃61;0.8.0 cudatoolkit&＃61;10.1 -c pytorch


大功告成&＃xff1a;