康行天下
你的分类损失比较大，看起来不太正常。和另一个issue里边的问题一样，我很长时间不用这份代码了，也不好找出原因是什么。或许你换个proto再试试，或者换到我commit里边的历史版本，不清楚是不是因为我做了哪些改动造成的。
@SuperPowerLF2 优化器类型也是在solver.proto里边指定，优化器的自定义可以在sgd_solver.cpp里边改。
@SuperPowerLF2 设置为0表示固定初始参数不变，不进行学习。
The script is in dir 'examples/YOLO'. I think the script also works for tiny yolo or needs some modification.
@claudehang Hi, you can remove the layer in yolov3.cfg file since it has no weight params. And yolo layer functions is implemented in test_yolov3.cpp
@claudehang I'm not sure what caused this problem. Maybe your `.weight` file is mismatch with `.cfg` or there is some params in conv layer that the script cannot process correctly....
@zjbaby I can't directly think of what the problem is, it may be related to your compilation environment.Make a clean copy and try again later.
see the QA part in README: > Cannot load module layer dynamic library, the program search the modules first in enviroment variable CAFFE_LAYER_PATH then in predefined DEFAULT_LAYER_PATH in Makefile. So...
@moyans There is no need run the `convert_model.py` since bbox unnormalize is done at test time in the C++ code.
@jinfagang Try lower the threshold, if there is no object, that may be caused by wrong labels used for training.