search

TheLegendAli / DeepLab-Context

Other
239 stars 97 forks source link

F1120 22:05:28.240000 34447 net.cpp:717] Check failed: target_blobs[j]->channels() == source_layer.blobs(j).channels() (1 vs. 3) Incompatible parameter size for layer conv1_1 #11

Closed makaspacex closed 1 year ago

makaspacex commented 7 years ago

Training1 net voc12/vgg128_noup Running ./build/tools/caffe train --solver=voc12/config/vgg128_noup/solver_train_aug.prototxt --weights=voc12/model/vgg128_noup/init.caffemodel --gpu=0 I1120 22:05:22.482954 34447 caffe.cpp:102] Use GPU with device ID 0 I1120 22:05:26.820761 34447 caffe.cpp:110] Starting Optimization I1120 22:05:26.820894 34447 solver.cpp:32] Initializing solver from parameters: train_net: "voc12/config/vgg128_noup/train_train_aug.prototxt" .... .... .... .... .... I1120 22:05:27.910971 34447 solver.cpp:41] Solver scaffolding done. I1120 22:05:27.910979 34447 caffe.cpp:118] Finetuning from voc12/model/vgg128_noup/init.caffemodel I1120 22:05:28.239959 34447 net.cpp:707] Copying source layer conv1_1 F1120 22:05:28.240000 34447 net.cpp:717] Check failed: target_blobs[j]->channels() == source_layer.blobs(j).channels() (1 vs. 3) Incompatible parameter size for layer conv1_1 Check failure stack trace: @ 0x7fe28b87aa5d google::LogMessage::Fail() @ 0x7fe28b87eef7 google::LogMessage::SendToLog() @ 0x7fe28b87cd59 google::LogMessage::Flush() @ 0x7fe28b87d05d google::LogMessageFatal::~LogMessageFatal() @ 0x5946d0 caffe::Net<>::CopyTrainedLayersFrom() @ 0x594df3 caffe::Net<>::CopyTrainedLayersFrom() @ 0x56c301 train() @ 0x56e88f main @ 0x30f8a1ed1d (unknown) @ 0x56b8a9 (unknown)

Process finished with exit code 0

please help me ,thanks

TheLegendAli commented 7 years ago

This means Deeplab expected an RGB input, but it looks like your source is grayscale

makaspacex commented 7 years ago

@TheLegendAli Thanks, the problem may be happened at data source

wolf943134497 commented 7 years ago

@zhuihunmiling @TheLegendAli I have similar error.Did you solve it?My data source download from VOC 2012 official website I0104 17:21:34.282826 31297 net.cpp:208] This network produces output accuracy I0104 17:21:34.282845 31297 net.cpp:467] Collecting Learning Rate and Weight Decay. I0104 17:21:34.282855 31297 net.cpp:219] Network initialization done. I0104 17:21:34.282858 31297 net.cpp:220] Memory required for data: 9170170576 I0104 17:21:34.282920 31297 solver.cpp:41] Solver scaffolding done. I0104 17:21:34.282927 31297 caffe.cpp:118] Finetuning from voc12/model/vgg128_noup/init.caffemodel I0104 17:21:34.386422 31297 net.cpp:740] Target layer fc8_voc12 not initialized. I0104 17:21:34.387146 31297 solver.cpp:160] Solving vgg128_noup I0104 17:21:34.387156 31297 solver.cpp:161] Learning Rate Policy: step E0104 17:21:34.854878 31306 io.cpp:76] Could not open or find file /home/gwm/VOCdevkit/VOC2012/SegmentationClassAug/2011_000541.png F0104 17:21:34.855020 31306 data_transformer.cpp:332] Check failed: img_height == seg_height (326 vs. 0) Check failure stack trace: @ 0x7f41d54d8daa (unknown) @ 0x7f41d54d8ce4 (unknown) @ 0x7f41d54d86e6 (unknown) @ 0x7f41d54db687 (unknown) @ 0x4dcc6e caffe::DataTransformer<>::TransformImgAndSeg() @ 0x59b88d caffe::ImageSegDataLayer<>::InternalThreadEntry() @ 0x7f41d3160829 thread_proxy @ 0x7f41d24f8184 start_thread @ 0x7f41d222537d (unknown) @ (nil) (unknown) Aborted (core dumped)