-
Notifications
You must be signed in to change notification settings - Fork 5
Caffe
Shuai YUAN edited this page Nov 4, 2019
·
27 revisions
- http://caffe.berkeleyvision.org/installation.html
- Caffe + Ubuntu 14.04 64bit + CUDA 6.5 配置说明
- Ubuntu 16.04 安装配置Caffe 图文详解
- Extract weights of network in caffe
- https://pastebin.com/nAGEV9EK
- Extracting weights from .caffemodel without caffe installed in Python
- Use CUDA 7's per-thread default stream, with automatic syncing #2077
- Synchronous SGD via layer-wise parallelism #2219
GLOG_minloglevel
- Setting GLOG_minloglevel=1 to prevent output in shell from Caffe
- Monitor training/validation process in Caffe
- 【神经网络与深度学习】【C/C++】C++日志操作开源函数库之Google-glog
- Google glog rolling files
- How To Use Google Logging Library (glog)
- https://github.com/googlesamples/assistant-sdk-python/issues/8
- https://github.com/tensorflow/tensorflow/issues/6341
make clean
ldconfig
make
- https://stackoverflow.com/questions/7816607/opencv-2-3-compiling-issue-undefined-refence-ubuntu-11-10
- https://github.com/BVLC/caffe/issues/3700
- http://answers.opencv.org/question/5498/undefined-reference-to-cvimreadstdstring-const-int/
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
- https://stackoverflow.com/questions/31307181/installation-issue-with-protobuf-and-caffe
- This program requires version 3.3.0 of the Protocol Buffer runtime library
- Protobuf 3.0.0 beta4 support
- make pycaffe error #50
- ImportError: No module named skimage.io #5716
- ImportError: No module named skimage.io #13
sudo ln /dev/null /dev/raw1394
- ctypes error: libdc1394 error: Failed to initialize libdc1394
- OpenCV: libdc1394 error: Failed to initialize libdc1394
Program hit cudaErrorCudartUnloading (error 29) due to "driver shutting down" on CUDA API call to cudaFree.
- I set device id not 0, but caffe always use gpu0 #6042
- CUDA Pro Tip: Control GPU Visibility with CUDA_VISIBLE_DEVICES
- would the instance of caffe running on gpu 1 with pythonlayer always take some memory of gpu 0? #5983
- GPU 0 is also used when running on other GPUs (#440 reocurred?) #2186
- GPU Memory Usage for Multiple GPUS #1399
- USE OpenCV 3
-Wl,--whole-archive libcaffe.a -Wl,--no-whole-archive