YOLOv3 conversion failed

[dsgh2]

Hi, I have a problem converting YOLOv3.

in darknet2caffe.py, Line20, net = caffe.Net(protofile, caffe.TEST)

net has not any values(=empty)

It seems to be worked like below

<_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> layer { <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> bottom: "layer105-conv" <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> top: "layer106-conv" <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> name: "layer106-conv" <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> type: "Convolution" <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> convolution_param { <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> num_output: 18 <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> kernel_size: 1 <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> pad: 0 <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> stride: 1 <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> bias_term: true <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> } <_io.TextIOWrapper name='result\\yolov3.prototxt' mode='w' encoding='cp949'> } however, it says I0724 00:32:51.248684 5636 net.cpp:255] Network initialization done. Traceback (most recent call last): File "darknet2caffe.py", line 522, in darknet2caffe(cfgfile, weightfile, protofile, caffemodel) File "darknet2caffe.py", line 64, in darknet2caffe start = load_conv_bn2caffe(buf, start, params[conv_layer_name], params[bn_layer_name], params[scale_layer_name]) KeyError: 'layer1-conv' How can I fix this? Thanks in advance

[sakethkhandavalli]

I0825 16:14:20.889731 26193 layer_factory.hpp:77] Creating layer input I0825 16:14:20.889760 26193 net.cpp:86] Creating Layer input I0825 16:14:20.889766 26193 net.cpp:382] input -> data I0825 16:14:20.889814 26193 net.cpp:124] Setting up input I0825 16:14:20.889819 26193 net.cpp:131] Top shape: 1 3 320 320 (307200) I0825 16:14:20.889824 26193 net.cpp:139] Memory required for data: 1228800 I0825 16:14:20.889827 26193 net.cpp:202] input does not need backward computation. I0825 16:14:20.889830 26193 net.cpp:244] This network produces output data I0825 16:14:20.889833 26193 net.cpp:257] Network initialization done. Traceback (most recent call last): File "darknet2caffe.py", line 519, in darknet2caffe(cfgfile, weightfile, protofile, caffemodel) File "darknet2caffe.py", line 61, in darknet2caffe start = load_conv_bn2caffe(buf, start, params[conv_layer_name], params[bn_layer_name], params[scale_layer_name]) KeyError: 'layer1-conv'

I am getting a similar error too, 'layer1-conv' key is missing, can anyone please tell me how to fix it. I am trying to convert a yolo-v3 model to caffe. @ChenYingpeng @becauseofAI @dsgh2

[KelvinHuang666]

i got the same bug when i changing darknet yolov2 model to caffe,how can I solve it,thanks.

[yousong1992]

same error KeyError:'layer1-conv' is missing

[yousong1992]

@sakethkhandavalli @ChenYingpeng plz help

[ChenYingpeng]

@yousong1992 @KelvinHuang666 @dsgh2 @sakethkhandavalli pls see this link.

[dsgh2]

Thank you!