Code issues - Githubissues

SeokjuLee commented 6 years ago

Please ask installation, training and test issues in this panel.

IAmNotMahd commented 6 years ago

@cardwing hey! I have been working on this for a while now. How do you display the output? In other words, what am I supposed to do after the last step (run "train.sh"). Thanks!

moonmyth commented 6 years ago

hello,i have a problem in training vpgnet,the loss of bbox is high while the accurancy is 95%,can anyone have idea that what's cause the phenomenon,use the dataset of caltech-lane-detection like blew: I0804 03:12:02.197446 6967 solver.cpp:414] Test net output #0: bb-loss = 1.17429 ( 1 = 1.17429 loss) I0804 03:12:02.197453 6967 solver.cpp:414] Test net output #1: pixel-acc = 0.967028 I0804 03:12:02.197458 6967 solver.cpp:414] Test net output #2: pixel-loss = 0.10597 ( 1 = 0.10597 loss) I0804 03:12:02.197461 6967 solver.cpp:414] Test net output #3: type-acc = 0.952496 I0804 03:12:02.197466 6967 solver.cpp:414] Test net output #4: type-loss = 0.220534 (* 1 = 0.220534 loss)

chenchaohui commented 5 years ago

Thanks for your guide.But I still have some question for the demos and tests. I use the command "./build/tools/caffe test -model models/vpgnet-novp/deploy.prototxt -weights models/vpgnet-novp/snapshots/split_iter_82500.caffemodel -iterations 1 >>output.log 2>&1",I want to know how to input a picture and get the cordinate of the lane?Because deploy.

Hi,I Also met this question ,can you tell me how to solve it?

chenchaohui commented 5 years ago

Hi, Please confirm whether you have set the environment variable of caffe you complied.

------------------ 原始邮件 ------------------ 发件人: "jiang4869"notifications@github.com; 发送时间: 2019年9月24日(星期二) 晚上6:08 收件人: "SeokjuLee/VPGNet"VPGNet@noreply.github.com; 抄送: "抑扬人"1433692592@qq.com;"Comment"comment@noreply.github.com; 主题: Re: [SeokjuLee/VPGNet] Code issues (#3)

Hello, when I downloaded the code from GitHub and compiled and tested the caffe, I encountered the following problems when running the train.sh file:

I0924 18:00:51.584237 31829 solver.cpp:102] Creating training net from net file:. /train_val.prototxt

[libprotobuf ERROR google/protobuf/text_format.cc:274] Error parsing Text-format caffe.NetParameter: 15:20: Message type "caffe.Layer Parameter" has no field named "drive_data_param".

F0924 18:00:51.584311 31829 upgrade_proto.cpp:90] Check failed: ReadProtoFromTextFile (param_file, param) Failed to parse NetParameter file: / train_val.prototxt.

What's the matter, please? Is the Caffe version used incompatible? I copy the files in the models directory to the models directory of my local compiled caffe.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

devyash17 commented 5 years ago

Hi @jiang4869 , try this sudo vim ~/.bashrc then add pycaffe to the PYTHONPATH by adding this line - export PYTHONPATH=/path/to/VPGNet/caffe/python:$PYTHONPATH

SeokjuLee / VPGNet

Code issues #3