win10, cuda 9.0, python 3.5环境下复现 Flow-Guided Feature Aggregation for Video Object Detection 问题总结demo训练测试

112 阅读 0 评论 74 点赞

我是靠谱客的博主粗犷超短裙，最近开发中收集的这篇文章主要介绍win10, cuda 9.0, python 3.5环境下复现 Flow-Guided Feature Aggregation for Video Object Detection 问题总结demo训练测试，觉得挺不错的，现在分享给大家，希望可以做个参考。

概述

demo

error:Microsoft Visual C++ is required
参考链接，在vs2017中配置vc++ 14
LINK : fatal error LNK1158: cannot run ‘rc.exe’
参考链接将rc.exe、rcdll.dll从C:Program Files (x86)Windows Kits8.1binx64复制到C:Program Files (x86)Microsoft Visual Studio 14.0VCbin
‘dict’ object has no attribute ‘iteritems’
对应文件的dict.iteritems改为dict.items （python 2 与 python 3的区别）
源代码使用python 2编写的，因此许多代码需要改动，主要为print、dict、pickle等函数。
LINK : fatal error LNK1181: 无法打开输入文件“ID=2.obj”

C:Program FilesNVIDIA GPU Computing ToolkitCUDAv9.0binnvcc.exe --compiler-bindir “C:Program Files (x86)Microsoft Visual Studio 14.0VCbin” --shared “-LD:Program FilesAnaconda3libs” “-LD:Program FilesAnaconda3PCbuildamd64” “-LC:Program Files (x86)Microsoft Visual Studio 14.0VCLIBamd64” “-LC:Program Files (x86)Microsoft Visual Studio 14.0VCATLMFCLIBamd64” “-LC:Program Files (x86)Windows Kits10lib10.0.17134.0ucrtx64” “-LC:Program Files (x86)Windows KitsNETFXSDK4.6.1libumx64” “-LC:Program Files (x86)Windows Kits10lib10.0.17134.0umx64” -lcublas buildtemp.win-amd64-3.5Releasegpu_nms.obj -o buildlib.win-amd64-3.5gpu_nms.cp35-win_amd64.pyd --linker-options=/nologo,/INCREMENTAL:NO,/LTCG,/MANIFEST:EMBED,ID=2,/MANIFESTUAC:NO,/IMPLIB:buildtemp.win-amd64-3.5Releasegpu_nms.cp35-win_amd64.lib,/NODEFAULTLIB:libcmt.lib
gpu_nms.obj
LINK : fatal error LNK1181: 无法打开输入文件“ID=2.obj”
error: command ‘C:Program FilesNVIDIA GPU Computing ToolkitCUDAv9.0binnvcc.exe’ failed with exit status 2

有一个曲线救国的方法：点此。出现改行错误后，将"ID=2,"去掉，手动执行该命令。有一个曲线救国的方法：点此。出现改行错误后，将"ID=2,"去掉，手动执行该命令。
mxnet.base.NotImplementedForSymbol: Function __iadd__ (namely operator "+=") with arguments (<class 'mxnet.symbol.symbol.Symbol'>, <class 'int'>) is not implemented for Symbol and only available in NDArray.
szha的回答
'.fgfa_rfcnsymbolsresnet_v1_101_flownet_rfcn.py’中的 aggregated_conv_feat += tiled_weight * warp_list[i]改为
aggregated_conv_feat = aggregated_conv_feat + tiled_weight * warp_list[i]
由于编译mxnet太麻烦，所以直接pip安装0.12.1版本；测试demo.py无问题。

训练

fail to open fgfa_rfcn_vid-0002.params

mxnet.base.MXNetError: [15:13:33] D:Program Files (x86)Jenkinsworkspacemxnetmxnetdmlc-coresrciolocal_filesys.cc:166: Check failed: allow_null LocalFileSystem: fail to open "./output/fgfa_rfcn/imagenet_vidresnet_v1_101_flownet_imagenet_vid_rfcn_end2end_ohemVID_val_videosDET_train_30classes_VID_train_15framesfgfa_rfcn_vid-0002.params"

注释掉experiments/fgfa_rfcn/fgfa_rfcn_end2end_train_test.py 的第20行。painterdrown的回答
TypeError: a bytes-like object is required, not ‘str’

Traceback (most recent call last):
File “_ctypes/callbacks.c”, line 234, in ‘calling callback function’
File “D:Program FilesAnaconda3libsite-packagesmxnetoperator.py”, line 621, in creator
op_prop = prop_cls(**kwargs)
File “Flow-Guided-Feature-Aggregation/experiments/fgfa_rfcn…fgfa_rfcnoperator_pyproposal_target.py”, line 94, in init
self._cfg = pickle.loads(cfg)
TypeError: a bytes-like object is required, not ‘str’

先将cfg用pickle.dumps序列化，再用pickle.loads反序列化，思路是没有问题的。但是不知为何，pickle.loads时cfg的格式变成了字符串而不是bytes——将值复制出来却又是bytes。索性将cfg序列化至磁盘，然后从磁盘中读出来。方法如下

fgfa_rfcn/symbols/resnet_v1_101_flownet_rfcn.py, line 965, 增加
```
with open('cfg.pck', 'wb') as fp:
pickle.dump(cfg, fp)
```
且原本的970行改为 cfg=None

fgfa_rfcnoperator_pyproposal_target.py, line 36、line 94, 改为
```
with open('cfg.pck', 'rb') as fp:
self._cfg = pickle.load(fp)
```
终于，成功进行训练。

Epoch[0] Batch [100] Speed: 0.44 samples/sec Train-RPNAcc=0.907294, RPNLogLoss=0.296552, RPNL1Loss=0.189087,
RCNNAcc=0.807704, RCNNLogLoss=1.681007, RCNNL1Loss=0.130364,
Epoch[0] Batch [200] Speed: 0.44 samples/sec Train-RPNAcc=0.923818, RPNLogLoss=0.242203, RPNL1Loss=0.156525,
RCNNAcc=0.861396, RCNNLogLoss=1.166992, RCNNL1Loss=0.163834,