refinedet记录

最新推荐文章于 2020-10-08 15:02:02 发布

原创最新推荐文章于 2020-10-08 15:02:02 发布 · 559 阅读

1 ·

CC 4.0 BY-SA版权

目标检测专栏收录该内容

1 篇文章

订阅专栏

!!!修改完Pascal_voc.py后记得重新编译生成对应的pyc

不然会一直执行未修改前的脚本

生成单个pyc文件

参考https://www.cnblogs.com/dkblog/archive/2009/04/16/1980757.html

对于py文件，可以执行下面命令来生成pyc文件。

python -m foo.py

另外一种方式是通过代码来生成pyc文件。

import py_compile
py_compile.compile('/path/to/foo.py')

!!!修改为二分类问题

https://blog.youkuaiyun.com/samylee/article/details/52860038

FPS

https://github.com/sfzhang15/RefineDet/issues/17

F0119 13:43:46.399646 5365 net.cpp:774] Cannot copy param 0 weights from layer 'P3_mbox_conf'; shape mismatch. Source param shape is 63 256 3 3 (145152); target param shape is 6 256 3 3 (13824). To learn this layer's parameters from scratch rather than copying from a saved net, rename the layer.
*** Check failure stack trace: ***
    @     0x7f60e2e8a5cd google::LogMessage::Fail()
    @     0x7f60e2e8c433 google::LogMessage::SendToLog()
    @     0x7f60e2e8a15b google::LogMessage::Flush()
    @     0x7f60e2e8ce1e google::LogMessageFatal::~LogMessageFatal()
    @     0x7f60e36dac8b caffe::Net<>::CopyTrainedLayersFrom()
    @     0x7f60e36e19e5 caffe::Net<>::CopyTrainedLayersFromBinaryProto()
    @     0x7f60e36e1a7e caffe::Net<>::CopyTrainedLayersFrom()
    @           0x40a869 CopyLayers()
    @           0x40bcc4 train()
    @           0x4077e8 main
    @     0x7f60e1620830 __libc_start_main
    @           0x4080b9 _start
    @              (nil) (unknown)
Aborted (core dumped)