labelme数据集标注制作culane格式数据集

0、MP4文件截图（saveframe.py）

import cv2
import os

cap = cv2.VideoCapture("re.mp4")      #xxx.mp4为文件名

a = os.getcwd()+'\\re'                #获取路径
os.mkdir(a)

frames_total = cap.get(7)
print('frames_total:',frames_total)
frames_need = 250                     #生成的图片数，需小于总数否则报错

print('frames_need:',frames_need)
n = 1
timeF = frames_total//frames_need
print('frames_gap:',timeF)
i = 0
while cap.isOpened():
    ret, frame = cap.read()
    if n%timeF == 0:
        i += 1
        print(i)
        cv2.imwrite('./re/re{}.jpg'.format(i),frame) #生成后的放置路径
        if i == frames_need:
            break
    n = n+1
    cv2.waitKey(1)
cap.release()

1、Anaconda下载创建labelme环境并下载labelme库

conda create -name labelme python=3.6

conda activate labelme 

pip install labelme 

2、 使用与运行

进入labelme环境后直接输入labelme进入标注程序

 file中设置自动保存

快捷键 ctrl+z撤销       a 上一张       d下一张

车道线使用create linestrip

3、Json格式文件批量转label图片

        在anaconda\envs\labelme\Lib\site-packages\labelme\cli中找到json_to_dataset.py制作备份后修改源文件

 

import argparse
import base64
import json
import os
import os.path as osp

import imgviz
import PIL.Image

from labelme.logger import logger
from labelme import utils


def main():
    logger.warning(
        "This script is aimed to demonstrate how to convert the "
        "JSON file to a single image dataset."
    )
    logger.warning(
        "It won't handle multiple JSON files to generate a "
        "real-use dataset."
    )

    parser = argparse.ArgumentParser()
    parser.add_argument("json_file")
    parser.add_argument("-o", "--out", default=None)
    args = parser.parse_args()

    json_file = args.json_file
    filename = json_file[:-5]
    if args.out is None:
        out_dir = osp.basename(json_file).replace(".", "_")
        print('_______',out_dir)
        filename = out_dir[:-5]
        out_dir = osp.join(osp.dirname(json_file), out_dir)
    else:
        out_dir = args.out
        
    print('*****************',filename)
   # if not osp.exists(out_dir):
     #   os.mkdir(out_dir)

    data = json.load(open(json_file))
    imageData = data.get("imageData")

    if not imageData:
        imagePath = os.path.join(os.path.dirname(json_file), data["imagePath"])
        with open(imagePath, "rb") as f:
            imageData = f.read()
            imageData = base64.b64encode(imageData).decode("utf-8")
    img = utils.img_b64_to_arr(imageData)

    label_name_to_value = {"_background_": 0}
    for shape in sorted(data["shapes"], key=lambda x: x["label"]):
        label_name = shape["label"]
        if label_name in label_name_to_value:
            label_value = label_name_to_value[label_name]
        else:
            label_value = len(label_name_to_value)
            label_name_to_value[label_name] = label_value
    lbl, _ = utils.shapes_to_label(
        img.shape, data["shapes"], label_name_to_value
    )

    label_names = [None] * (max(label_name_to_value.values()) + 1)
    for name, value in label_name_to_value.items():
        label_names[value] = name

    lbl_viz = imgviz.label2rgb(
        lbl, imgviz.asgray(img), label_names=label_names, loc="rb"
    )
    print('22222',out_dir,type(out_dir))
   # out_dir = out_dir[:12]
    #PIL.Image.fromarray(img).save(osp.join(out_dir, "img.png"))
    utils.lblsave(osp.join( out_dir+".png"), lbl)
    #PIL.Image.fromarray(lbl_viz).save(osp.join(out_dir, "label_viz.png"))

   # with open(osp.join(out_dir, "label_names.txt"), "w") as f:
       # for lbl_name in label_names:
           # f.write(lbl_name + "\n")

    logger.info("Saved to: {}".format(out_dir))


if __name__ == "__main__":
    main()

然后新建批量转化函数create.py(与标注好的JSON 文件放置于同一目录下):

import os
#json_folder = r" C:\Users\Jason\Desktop\dataset\re_DJI_20201002103929_0033_S"
json_folder = os.getcwd()
# 获取文件夹内的文件名
FileNameList = os.listdir(json_folder)
# 激活labelme环境
os.system("conda activate labelme")
for i in range(len(FileNameList)):
# 判断当前文件是否为json文件
    if(os.path.splitext(FileNameList[i])[1] == ".json"):
        json_file = json_folder + "\\" + FileNameList[i]
# 将该json文件转为png
        os.system("labelme_json_to_dataset " + json_file)

在labelme环境下CD至该目录下运行，完成3文件（如下图）

4、 Culane格式数据集制作 

1) 新建 lanesuidao文件夹，其中新建all文件夹和list文件夹

2) jpg 和 json分开放置在新的文件夹中(为了分别使用程序形成TXT文件)

 

3) trainlist.py生成trainlist.txt,其中改三处地方，文件路径、txt名字和file.write里的名字(要和第一步的 all相同 /all/）

#!/usr/bin/python
import os
def ListFilesToTxt(dir,file,wildcard,recursion):
    exts = wildcard.split(" ")
    files = os.listdir(dir)
    for name in files:
        fullname=os.path.join(dir,name)
        if(os.path.isdir(fullname) & recursion):
            ListFilesToTxt(fullname,file,wildcard,recursion)
        else:
            for ext in exts:
                if(name.endswith(ext)):
                    file.write("/all/"+name + "\n") #存放合集（原图和json等）文件夹的名字
                    break
def Test():
    dir="C:\Users\Jason\Desktop\dataset\gaokong\yuantu" #存放原图文件夹(只存放原图)路径
    outfile="trainlist.txt"                                  #写入的txt文件名
    wildcard = ".jpg"                                      #要读取的文件类型；
    file = open(outfile,"w")
    if not file:
        print ("cannot open the file %s for writing" % outfile)
    ListFilesToTxt(dir,file,wildcard, 1)
    file.close()


Test()

4) json2txt.py生成txt文件 新建list_linshi文件夹(每个jpg应一个txt)，json2txt.py修改两处路径

import os
import json
import numpy as np
dir_json = 'C:\Users\Jason\Desktop\dataset\gaokong\json'   #json存储的文件目录
dir_txt = 'C:\Users\Jason\Desktop\dataset\gaokong\list_linshi'     #txt存储目录
if not os.path.exists(dir_txt):
    os.makedirs(dir_txt)
list_json = os.listdir(dir_json)
def json2txt(path_json,path_txt):           #可修改生成格式
    with open(path_json,'r') as path_json:
        jsonx=json.load(path_json)
        with open(path_txt,'w+') as ftxt:
            for shape in jsonx['shapes']:
                label = str(shape['label'])+' '
                xy=np.array(shape['points'])
                strxy = ''

                for m,n in xy:
                    m=int(m)
                    n=int(n)
                    # print('m:',m)
                    # print('n：',n)
                    strxy+=str(m)+' '+str(n)+' '
                
                label = strxy
                ftxt.writelines(label+"\n")
for cnt,json_name in enumerate(list_json):
    print('cnt=%d,name=%s'%(cnt,json_name))
    path_json = dir_json + json_name
    print(path_json)
    path_txt = dir_txt + json_name.replace('.json','.lines.txt')
    print(path_txt)
    json2txt(path_json,path_txt)

5) 将隧道的原图 和 上述所有的txt文件拷贝到 lane/yuantu/

culane 数据集由jpg和相应的txt文件组成

6) 将trainlist.txt复制到 lanesuidao/list/

7) 进入seg_label_generate 文件夹，修改labelGen.sh中的路径为四个文件所在目录的路径,

生成laneseg_label

原本来源：github

稍作修改分开函数:

#!/bin/bash

# modify CULane to yours
CULane=/home/tt/mfz/lane/lane/data/gaokong  #修改
OutputPath=${CULane}/laneseg_label
if [ ! -d $OutputPath ]; then
  mkdir $OutputPath
fi
./seg_label_generate \
    -l ${CULane}/list/train.txt \
    -m imgLabel \
    -d $CULane \
    -w 16 \
    -o $OutputPath \

# explanation:
# -l: image list file to process
# -m: set mode to "imgLabel" or "trainList"
# -d: dataset path
# -w: the width of lane labels generated
# -o: path to save the generated labels
# -s: visualize annotation, remove this option to generate labels

8) 进入某一个anaconda虚拟环境，（包含Opencv）

sh labelGen.sh,可能会提示opencv相关的错误，

解决:sudo apt-get install libopencv-dev，再sh 部分问题可见博客

9) 将laneseg_label里的label图都删掉，将在自己的PNG图替换到这儿。

注意名字需对应不可带_josn，如带后缀需批量改名

10) 将labelListGen.sh里的路径修改，sh labellistGen.sh生成train_gt.txt

修改后代码：

#!/bin/bash

# modify CULane to yours
CULane=/home/tt/mfz/lane/lane/data/gaokong #修改
OutputPath=${CULane}/laneseg_label
if [ ! -d $OutputPath ]; then
  mkdir $OutputPath
fi
./seg_label_generate \
    -l ${CULane}/list/train.txt \
    -m trainList \
    -d $CULane \
    -w 16 \
    -o $OutputPath \

# explanation:
# -l: image list file to process
# -m: set mode to "imgLabel" or "trainList"
# -d: dataset path
# -w: the width of lane labels generated
# -o: path to save the generated labels
# -s: visualize annotation, remove this option to generate labels

三个文件夹构成数据集

yuantu (原图jpg和相应的json2txt.py生成的txt文件组成)

laneseg_label(嵌套文件夹下包含替换后的PNG)

list(包含俩个txt文件，trainlist.py生成的train.txt     和  labelListGen.sh生成的train_gt.txt)

11) 修改configs/culane.py中的相关参数 和 路径可以开始训练 注意 数据集文件夹 log 代码文件夹不 要放在一起**