Caffe - yszheda/wiki GitHub Wiki

Installation

Extract weight

pycaffe

mini caffe

MemPool

Performance

multi-gpu

multi-thread

multi CUDA Stream

log

GLOG_minloglevel

google protobuf

from google import protobuf ImportError: No module named google

Trouble-shooting

[Build] undefined reference to cv::imread...

make clean
ldconfig
make

[Build] make: protoc: Command not found

[Build] can't find hdf5.h

Makefile.config:

INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial/

Makefile:

--- LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_hl hdf5
+++ LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial

[Build] This program requires version 3.4.0 of the Protocol Buffer runtime library, but the installed version is 2.5.0. Please update your library.

源码编译protobuf

[Build] Python

ImportError: No module named skimage.io

libdc1394 error: Failed to initialize libdc1394

sudo ln /dev/null /dev/raw1394

cannot find -lopenblas

Check failed: status == CUDNN_STATUS_SUCCESS (8 vs. 0) CUDNN_STATUS_EXECUTION_FAILED

Program hit cudaErrorCudartUnloading (error 29) due to "driver shutting down" on CUDA API call to cudaFree.

CUDA: invalid device ordinal

gpu-0

classification.o: undefined reference to symbol '_ZN2cv6imreadERKNS_6StringEi'

  • USE OpenCV 3

static library

Check failed: registry.count(type) == 1 (0 vs. 1) Unknown layer type: Input (known types: )

-Wl,--whole-archive libcaffe.a -Wl,--no-whole-archive

CMake

A protocol message was rejected because it was too big (more than 67108864 bytes). To increase the limit (or to disable these warnings), see CodedInputStream::SetTotalBytesLimit() in google/protobuf/io/coded_stream.h.