frcnn_roi_data_layer.cpp:220] Check failed: rois[i][DataPrepare::X2] < cols (500 vs. 500) rois[i][DataPrepare::X2] : 500

[moyans]

No description provided.

[moyans]

I1130 15:56:00.993705 7514 solver.cpp:231] Iteration 0 (-5.23529 iter/s, 0.543508s/100 iters), loss = 5.15557 [ 0 / 12000 ] -> [ 0:0 (H:M) ]
I1130 15:56:00.993734 7514 solver.cpp:257] Train net output #0: bbox_accuracy = 0.0078125
I1130 15:56:00.993741 7514 solver.cpp:257] Train net output #1: loss_bbox = 0.679996 (* 2 = 1.35999 loss)
I1130 15:56:00.993746 7514 solver.cpp:257] Train net output #2: loss_cls = 3.0499 (* 1 = 3.0499 loss)
I1130 15:56:00.993749 7514 solver.cpp:257] Train net output #3: rpn_cls_loss = 0.719319 (* 1 = 0.719319 loss)
I1130 15:56:00.993754 7514 solver.cpp:257] Train net output #4: rpn_loss_bbox = 0.207151 (* 1 = 0.207151 loss)
I1130 15:56:00.993760 7514 sgd_solver.cpp:148] Iteration 0, lr = 0.00033
F1130 15:56:01.850818 7526 frcnn_roi_data_layer.cpp:220] Check failed: rois[i][DataPrepare::X2] < cols (500 vs. 500) rois[i][DataPrepare::X2] : 500
*** Check failure stack trace: ***
@ 0x7fe1d750b5cd google::LogMessage::Fail()
@ 0x7fe1d750d433 google::LogMessage::SendToLog()
@ 0x7fe1d750b15b google::LogMessage::Flush()
@ 0x7fe1d750de1e google::LogMessageFatal::~LogMessageFatal()
@ 0x7fe1d7f6fcf5 caffe::Frcnn::FrcnnRoiDataLayer<>::FlipRois()
@ 0x7fe1d7f72851 caffe::Frcnn::FrcnnRoiDataLayer<>::load_batch()
@ 0x7fe1d7e55433 caffe::BasePrefetchingDataLayer<>::InternalThreadEntry()
@ 0x7fe1d7dbb0b5 caffe::InternalThread::entry()
@ 0x7fe1d7dbbf7e boost::detail::thread_data<>::run()
@ 0x7fe1cdb8c5d5 (unknown)
@ 0x7fe1c835e6ba start_thread
@ 0x7fe1d656f41d clone
@ (nil) (unknown)
已放弃 (核心已转储)

[makefile]

You can modify your annotation file make sure that x2 is less than image width. Or change the code at frcnn_roi_data_layer.cpp#220 from CHECK_LT to CHECK_LE. the coordinates x,y start from 0 instead of 1 (the PASCAL VOC start from 1).

[moyans]

The problem has been solved, thks.