美文网首页
Feature Pyramid Networks for Obj

Feature Pyramid Networks for Obj

作者: 晓智AI | 来源:发表于2019-03-23 20:36 被阅读0次

研究背景

准备进行行星/恒星等小目标检测

研究参考

论文链接
论文解读参考
Blog参考

算法编译

代码1链接
代码2链接

环境搭建

# packages in environment at /home/ouc/miniconda3/envs/fpn2:
#
# Name                    Version                   Build  Channel
blas                      1.0                         mkl    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
ca-certificates           2019.1.23                     0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
certifi                   2019.3.9                 py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
cffi                      1.12.2           py27h2e261b9_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
cudatoolkit               8.0                           3    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free
cudnn                     7.1.3                 cuda8.0_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
freetype                  2.9.1                h8a8886c_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
intel-openmp              2019.3                      199    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
jpeg                      9b                   h024ee3a_2    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libedit                   3.1.20181209         hc058e9b_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libffi                    3.2.1                hd88cf55_4    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libgcc-ng                 8.2.0                hdf63c60_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libgfortran-ng            7.3.0                hdf63c60_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libpng                    1.6.36               hbc83047_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libstdcxx-ng              8.2.0                hdf63c60_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libtiff                   4.0.10               h2733197_2    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
mkl                       2018.0.3                      1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
mkl_fft                   1.0.6            py27h7dd41cf_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
mkl_random                1.0.1            py27h4414c95_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
nccl                      1.3.4                 cuda8.0_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free
ncurses                   6.1                  he6710b0_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
ninja                     1.8.2            py27h6bb024c_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
numpy                     1.15.4           py27h1d66e8a_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
numpy-base                1.15.4           py27h81de0dd_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
olefile                   0.46                     py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
openssl                   1.1.1b               h7b6447c_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pillow                    5.4.1            py27h34e0f95_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pip                       19.0.3                   py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pycparser                 2.19                     py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
python                    2.7.16               h9bab390_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pytorch                   0.3.1            py27hfbe7015_1    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pyyaml                    5.1              py27h7b6447c_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
readline                  7.0                  h7b6447c_5    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
setuptools                40.8.0                   py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
six                       1.12.0                   py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
sqlite                    3.27.2               h7b6447c_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
tk                        8.6.8                hbc83047_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
torchvision               0.2.0                    py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
wheel                     0.33.1                   py27_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
xz                        5.2.4                h14c3975_4    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
yaml                      0.1.7                had09818_2    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
zlib                      1.2.11               h7b6447c_3    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
zstd                      1.3.7                h0b5b093_0    https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main

训练前准备

  1. 数据集
    VOC2007
    存放路径 /data/VOCdevkit2007/VOC2007

  2. sh make.sh
    报错


src/binding.cpp:6:10: fatal error: torch/extension.h: No such file or directory

Pytorch版本问题
解决fatal error: torch/extension.h: No such file or directory

报错

distutils.errors.DistutilsExecError: command 'gcc' failed with exit status 1

PANet训练自己的数据_make报错解决

训练报错

  1. indices报错
TypeError: index(): argument 'indices' (position 1) must be tuple of Tensors, not Tensor in proposal_target_layer_cascade.py

解决

labels = gt_boxes[:, :, 4].contiguous().view(-1).index((offset.view(-1),)).view(batch_size, -1)`

参考

  1. proposals_single 报错
#output[i,:num_proposal,1:] = proposals_single
if num_proposal == 0:
     num_proposal = num_proposal + 1
     proposals_single = torch.zeros((1, 4))
     output[i, :num_proposal, 1:] = proposals_single

参考

  1. 报错 RCNN_rpn was not defined
RCNN_rpn.rpn_loss_cls.data[0]->
rpn_loss_cls.data[0]

参考

测试报错

  1. RuntimeError报错
    File "/home/ouc/data1/liuhongzhi/FPN/test_net.py", line 284, in <module>
    cls_dets = torch.cat((cls_boxes, cls_scores), 1)
    RuntimeError: invalid argument 0: Tensors must have same number of dimensions: got 1 and 2 at /opt/conda/conda-bld/pytorch_1525796793591/work/torch/lib/THC/generic/THCTensorMath.cu:102

解决:

#cls_dets = torch.cat((cls_boxes, cls_scores), 1)

cls_dets = torch.cat((cls_boxes, cls_scores.unsqueeze(1)), 1)

知识补充

  1. 计算模型以及中间变量的显存占用大小
    浅谈深度学习:如何计算模型以及中间变量的显存占用大小

  2. pytorch 减小显存消耗,优化显存使用
    pytorch 减小显存消耗,优化显存使用,避免out of memory

  3. git clone 子模块下载不全问题
    有时含有子模块的git库下载不全,比如


git clone https://github.com/mahyarnajibi/SNIPER.git

此时应输入如下指令

git clone --recursive https://github.com/mahyarnajibi/SNIPER.git

参考链接

相关文章

网友评论

      本文标题:Feature Pyramid Networks for Obj

      本文链接:https://www.haomeiwen.com/subject/kxdsvqtx.html