Hi,
I sucessfully was able to set up caffe in an Ubuntu 16.04 PC, I did it in the CPU mode (Makefile.config)
Running train_image_object_detection.sh results in two runtime errors:
1.
F0204 14:53:48.231036 22981 benchmark.cpp:114] Cannot use GPU in CPU-only Caffe: check mode.
*** Check failure stack trace: ***
@ 0x7f89b99af5cd google::LogMessage::Fail()
@ 0x7f89b99b1433 google::LogMessage::SendToLog()
@ 0x7f89b99af15b google::LogMessage::Flush()
@ 0x7f89b99b1e1e google::LogMessageFatal::~LogMessageFatal()
@ 0x7f89ba29f697 caffe::Timer::Init()
@ 0x7f89ba2f69bd caffe::Solver::Solver()
@ 0x7f89ba25823e caffe::Creator_SGDSolver()
@ 0x4198a6 caffe::SolverRegistry::CreateSolver()
@ 0x4115d4 train()
@ 0x40c3eb main
@ 0x7f89b7fa4830 __libc_start_main
@ 0x40d009 _start
@ (nil) (unknown)
Does train_image_object_detection.sh have a cpu mode switch? There is gpus="0,1" variable but setting it to gpus="" just causes more problems at the Python side.
2.
F0204 14:53:48.969892 23028 db_lmdb.hpp:16] Check failed: mdb_status == 0 (2 vs. 0) No such file or directory
*** Check failure stack trace: ***
@ 0x7f45925155cd google::LogMessage::Fail()
@ 0x7f4592517433 google::LogMessage::SendToLog()
@ 0x7f459251515b google::LogMessage::Flush()
@ 0x7f4592517e1e google::LogMessageFatal::~LogMessageFatal()
@ 0x7f4592e31ae0 caffe::db::LMDB::Open()
@ 0x7f4592da72ea caffe::DataReader<>::InternalThreadEntryN()
@ 0x7f4592dec48c caffe::InternalThread::entry()
@ 0x7f4592dee33b boost::detail::thread_data<>::run()
@ 0x7f458f2a25d5 (unknown)
@ 0x7f458c9d66ba start_thread
@ 0x7f4590bf141d clone
@ (nil) (unknown)
This is, as far as I understood, related to the script not being able to find lmdb files, and they're not there.
How do I get those files to try out the default example from caffe-jacinto for SSD network training.
Thank you in advance,
Cheers,
