路径 /home/nvidia/tegra_multimedia_api/samples/11_camera_object_identification/

参考该路径下README 完成安装编译等操作。

使用sudo命令 切换至root用户操作，其中$HOME变量 替换为实际路径 /home/nvidia/ 如：文中58行 cd $HOME/Work/caffe/caffe-master

实际终端输入命令：cd /home/nvidia/Work/caffe/caffe-master

95行执行 ./scripts/download_model_binary.py  models/bvlc_reference_caffenet/

错误No module named six.moves

处理办，拷贝文件 cp /usr/lib/python2.7/dist-packages/pkg_resources/_vendor/six.py* ./scripts/ 

 如图1-1

图1-1 No module named six.moves错误处理

make -j4问题处理

../Rules.mk:37: *** Please specify which ARM ABI platform you are compiling for.  Stop.

export TEGRA_ARMABI=aarch64-linux-gnu

make clean;make

图1-2 声明TEGRA_ARMABI

升级最新pip

pip install --upgrade pip

8.1.1 升级到9.0.1

环境变量

vim ~/.bashrc

在~/.bashrc最后一行如下内容，保存并退出(wq)：

export TEGRA_ARMABI=aarch64-linux-gnu

export DISPLAY=:0

export LD_LIBRARY_PATH=/home/nvidia/caffe/caffe-master/build/lib:/usr/local/cuda/lib64

运行

./camera_caffe -width 1920 -height 1080 -lib opencv_consumer_lib/libopencv_consumer.so -model /home/nvidia/Work/caffe/caffe-master/models/bvlc_reference_caffenet/deploy.prototxt -trained /home/nvidia/Work/caffe/caffe-master/models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel -mean /home/nvidia/Work/caffe/caffe-master/data/ilsvrc12/imagenet_mean.binaryproto -label /home/nvidia/Work/caffe/caffe-master/data/ilsvrc12/synset_words.txt

可参考博文：

http://blog.csdn.net/c406495762/article/details/71116735