caffe之操蛋的create_imagenet

来源：互联网发布：程序员必读书单颈椎病编辑：程序博客网时间：2024/04/28 17:19

QAQ,鼓捣了三四天,反正我蠢我bb

train0.JPEG 0train1.JPEG 0train2.JPEG 0train3.JPEG 0train4.JPEG 0train5.JPEG 0train6.JPEG 0train7.JPEG 0......

train.txt应该这样写

cdcd 文档/sh_learnbash val.shbash train.sh

#! /bin/bashcdcd 文档/sh_learnpath=train2rm -r trainrm train.txtcleardir=$(ls $path)num_pic=0num_num=0num_name=0max=200  mkdir traindir2=$(ls $path$i)for i in $dir2doecho "数字:($i),编号:($num_num)"num_pic=0dir3=$(ls $path/$i)for j in $dir3doif [ $max -gt $num_pic ]thenecho "第($num_pic)张图片名($j),新图片名($num_name.jpg)"cp $path/$i/$j train/train$num_name.JPEGecho train$num_name.JPEG $num_num >> train.txt filet num_pic+=1let num_name+=1done let num_num+=1done

#! /bin/bashcdcd 文档/sh_learnpath=val2rm -r valrm val.txtcleardir=$(ls $path)num_pic=0num_num=0num_name=0max=200  mkdir valdir2=$(ls $path$i)for i in $dir2doecho "数字:($i),编号:($num_num)"num_pic=0dir3=$(ls $path/$i)for j in $dir3doif [ $max -gt $num_pic ]thenecho "第($num_pic)张图片名($j),新图片名($num_name.jpg)"cp $path/$i/$j val/train$num_name.JPEGecho train$num_name.JPEG $num_num >> val.txt filet num_pic+=1let num_name+=1done let num_num+=1done

这是我把图片放在一起并生成txt的

然后是那个该死的create_imagenet<oh,不是该死的,是我蠢=- =,QAQ>

（注意：这里用的是灰度图像，灰度灰度灰度gray）

#!/usr/bin/env shclearsudo rm -r examples/my_imagenet/ilsvrc12_train_lmdbsudo rm -r examples/my_imagenet/ilsvrc12_val_lmdbEXAMPLE=examples/my_imagenetDATA=data/my_ilsvrc12data/TOOLS=build/tools TRAIN_DATA_ROOT=data/my_ilsvrc12data/train/RESIZE=trueif $RESIZE; then RESIZE_HEIGHT=32 RESIZE_WIDTH=32else RESIZE_HEIGHT=0 RESIZE_WIDTH=0fiif [ ! -d "$TRAIN_DATA_ROOT" ];then echo "Error: TRAIN_DATA_ROOT is not a path to a directory:$TRAIN_DATA_ROOT" echo "Set the TRAIN_DATA_ROOT variable in create_imagenet.sh to thepath" \ "where the ImageNet training data is stored." exit 1fiecho "Creating train lmdb..."GLOG_logtostderr=1 $TOOLS/convert_imageset \ --resize_height=$RESIZE_HEIGHT \ --resize_width=$RESIZE_WIDTH \ --backend="lmdb" \ --gray=true \ --shuffle \ $TRAIN_DATA_ROOT \ $DATA/train.txt \ $EXAMPLE/ilsvrc12_train_lmdbVAL_DATA_ROOT=data/my_ilsvrc12data/val/if [ ! -d "$VAL_DATA_ROOT" ]; then echo "Error: VAL_DATA_ROOT is not a path to a directory:$VAL_DATA_ROOT" echo "Set the VAL_DATA_ROOT variable in create_imagenet.sh to thepath" \ "where the ImageNet validation data is stored." exit 1fiecho "Creating val lmdb..." GLOG_logtostderr=1 $TOOLS/convert_imageset \ --resize_height=$RESIZE_HEIGHT \ --resize_width=$RESIZE_WIDTH \

   --backend="lmdb" \   --gray=true \

--shuffle \ $VAL_DATA_ROOT \ $DATA/val.txt \ $EXAMPLE/ilsvrc12_val_lmdb echo "Done."

重点啊:

sh create_imagenet.sh

结果啊:

Creating train lmdb...I1031 10:05:10.154927 11937 convert_imageset.cpp:86] Shuffling dataI1031 10:05:10.155725 11937 convert_imageset.cpp:89] A total of 2000 images.I1031 10:05:10.155937 11937 db_lmdb.cpp:35] Opened lmdb examples/my_imagenet/ilsvrc12_train_lmdbI1031 10:05:11.350745 11937 convert_imageset.cpp:147] Processed 1000 files.I1031 10:05:12.523094 11937 convert_imageset.cpp:147] Processed 2000 files.Creating val lmdb...I1031 10:05:12.550520 11938 convert_imageset.cpp:86] Shuffling dataI1031 10:05:12.551501 11938 convert_imageset.cpp:89] A total of 2000 images.I1031 10:05:12.551832 11938 db_lmdb.cpp:35] Opened lmdb examples/my_imagenet/ilsvrc12_val_lmdbI1031 10:05:15.412485 11938 convert_imageset.cpp:147] Processed 1000 files.I1031 10:05:18.186136 11938 convert_imageset.cpp:147] Processed 2000 files.Done.erkp@erkp:~/caffe-master$ sudo bash '/home/erkp/caffe-master/data/my_ilsvrc12data/create_imagenet.sh'

结果是这个鸟样:

may force be with me

    --backend="lmdb" \    --gray=true \

1 0