本文介绍如何使用 Caffe DSSD 框架为特定的数据集进行预处理，包括设置工作目录、定义数据路径及名称、配置数据转换参数，并通过 Python 脚本创建标注文件。
root_dir=/home/jin/caffe_dssd
echo rootdircdrootdircdroot_dir
redo=1
data_root_dir=”$HOME/Downloads/PyData/Cares”
dataset_name=”VOCar_plate”
mapfile=”/home/jin/Downloads/PyData/Cares/labelmap_carplate.prototxt”
anno_type=”detection”
db=”lmdb”
min_dim=0
max_dim=0
width=0
height=0
extra_cmd=”–encode-type=jpg –encoded”
if [ redo]thenextracmd=”redo]thenextracmd=”extra_cmd –redo”
fi
for subset in test trainval
python rootdir/scripts/createannoset.py−−anno−type=rootdir/scripts/createannoset.py−−anno−type=anno_type
–label-map-file=mapfile−−min−dim=mapfile−−min−dim=min_dim –max-dim=maxdim–resize−width=maxdim–resize−width=width --resize-height=height–check−labelheight–check−labelextra_cmd datarootdirdatarootdirdata_root_dir/subset.txtsubset.txtdata_root_dir/datasetname/datasetname/db/datasetname””datasetname””subset"_"dbexamples/dbexamples/dataset_name
done
扫一扫
1543
到【灌水乐园】发言
