docker训练caffe的SSD

硬件配置:GTX 1080,

docker 的镜像配置是CUDA8,,在哪里看到说gtx1080不能用CUDA7.5及以下,todo:找出那个链接

镜像地址为docker pull xiaoxinyi/caffe-gpu:8.23

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~SSD训练

1.git clone https://github.com/weiliu89/caffe

cd caffe之后,git checkout ssd

2.然后make -j8 && make pycaffe

会出现两个错误:

(1)解决caffe make报错:/usr/include/boost/property_tree/detail/json_parser_read.hpp:257:264: error: ‘type name’ declared as function returning an array:

解决方法:https://blog.csdn.net/allyli0022/article/details/62881238

(2)安装sudo apt-get install libatlas-base-dev(不然编译caffe时候出错/usr/bin/ld:cannot find -lcblas和latlas

3.放VOC07格式数据集,并修改文件按照https://blog.csdn.net/lilai619/article/details/53791420

(1)预训练模型:链接: https://pan.baidu.com/s/1IzLh-YcW52tseZUldcEEhg 提取码: rprn 

(2)修改create_list.sh和create_data的时候,$HOME改成caffe所在路径

(2)在score_ssd_pascal.py/ssd_pascal_webcam.py/ssd_pascal_video.py等对应脚本中添加                sys.path.insert(0,'/home/xxx/caffe/python'),,,,否则create_data.sh的时候报错放

(3.按3.链接继续修改,运行:python ./examples/ssd/ssd_pascal.py 2>&1 | tee ssd_train_log.txt 开始训练

------------------------------------------------------------------------------------------------

参数:

学习率base_lr,  batch_size ,max_iteration, snapshot(保存间隔)

save_dir(模型保存路径)

=================================

第二步的报上有人说升级gcc,升级后还是报错gcc version later than 5.3 are not supported;不过按上面的链接修改/usr/include/boost/property_tree/detail/json_parser_read.hpp就可以了

 

你可能感兴趣的:(经验)