123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430 |
- // 1. 读取指定目录文件, 获取文件内容.
- // 2. 解析内容, 获取图片的标注信息
- // 3. 统计标注信息
- const fs = require('fs');
- const path = require('path');
- const readline = require('readline');
- function readFile(filePath) {
- return new Promise((resolve, reject) => {
- fs.readFile(filePath, 'utf-8', (err, data) => {
- if (err) return reject(err);
- resolve(data);
- });
- })
- }
- function readDir(filePath) {
- return new Promise((resolve, reject) => {
- fs.readdir(filePath, (err, files) => {
- if (err) return reject(err);
- resolve(files);
- });
- })
- }
- function parseYolo(fileData) {
- // 解析文件内容, 获取图片的标注信息
- // 类型 x坐标 y坐标 宽度 高度
- // 1 0.391297 0.095892 0.280578 0.179688
- let result = [];
- let lines = fileData.split('\n');
- lines.forEach(line => {
- // 判断是否为空行
- if (line.trim() === '') return;
- let [type, x, y, width, height] = line.split(' ');
- result.push({ type, x, y, width, height });
- });
- return result;
- }
- async function writeFile(filePath, fileData) {
- return new Promise((resolve, reject) => {
- // 判断父级目录是否存在
- let dirPath = path.dirname(filePath);
- if (!fs.existsSync(dirPath)) {
- fs.mkdirSync(dirPath, {recursive: true });
- }
- fs.writeFile(filePath, fileData, (err, data) => {
- if (err) return reject(err);
- resolve(data);
- });
- })
- }
- async function mvFile(filePath, targetPath) {
- return new Promise((resolve, reject) => {
- // 判断父级目录是否存在
- let dirPath = path.dirname(filePath);
- if (!fs.existsSync(dirPath)) {
- fs.mkdirSync(dirPath, {recursive: true });
- }
- fs.rename(filePath, targetPath, (err, data) => {
- if (err) return reject(err);
- resolve(data);
- });
- })
- }
- async function cpFile(filePath, targetPath) {
- return new Promise((resolve, reject) => {
- // 判断父级目录是否存在
- let dirPath = path.dirname(targetPath);
- if (!fs.existsSync(dirPath)) {
- console.log(`路径${dirPath} 不存在`)
- fs.mkdirSync(dirPath, {recursive: true });
- console.log('路径不存在')
- }
- // console.log('路径不存在')
- fs.copyFile(filePath, targetPath, (err, data) => {
- if (err) return reject(err);
- resolve(data);
- });
- })
- }
- // 删除指定目录
- function rmDir(path) {
- return new Promise((resolve, reject) => {
- fs.rmdir(path, { recursive: true }, (err, data) => {
- if (err) return reject(err);
- resolve(data);
- });
- })
- }
- const dimensionType = {
- "-1": "无标注",
- "0": "吊车",
- "1": "塔吊",
- "2": "烟火",
- "3": "施工机械",
- "4": "导线异物",
- "5": "烟雾"
- }
- async function main(){
- // let labelsPath = `E:\\图库\\ai\\epower_v2\\newLabels`
- // let labelsPath = `E:\\图库\\ai\\epower_v2\\处理labels`
- // let labelsPath = `E:\\图库\\ai\\epower_v2\\labels`
- let labelsPath = `E:\\图库\\ai\\epower\\all_tmpLabels_2`
- let imagesPath = `E:\\图库\\ai\\epower\\images`
- let emptyPath = `E:\\图库\\ai\\epower\\empty`
- let imageExts = ['jpg']
- let logsPath = `E:\\图库\\ai\\epower\\logs`
- // 数据转移目录
- let transferPath = `E:\\图库\\ai\\epower\\parse`
- // 清除数据转移目录
- // 判断数据转移目录是否存在
- if (fs.existsSync(transferPath)) { // 删除数据转移目录
- await rmDir(transferPath);
- }
- let countMap = {};
- let logStrArr = [];
- let resultStrArr = [];
- // 获取文件列表
- let labelFiles = await readDir(labelsPath);
- let imagesFiles = await readDir(imagesPath);
- // 获取当前时间
- let now = new Date();
- let nowStr = `${now.getFullYear()}-${now.getMonth()+1}-${now.getDate()} ${now.getHours()}点${now.getMinutes()}分${now.getSeconds()}秒`;
- resultStrArr.push(`[I] [START] 开始统计分析数据标注信息 ${nowStr}`);
- resultStrArr.push(`[I] 原标注信息路径: ${labelsPath}`);
- resultStrArr.push(`[I] 原图片文件路径: ${labelsPath}`);
- resultStrArr.push(`[I] 输出文件路径: ${labelsPath}`);
- resultStrArr.push(`[I] 输出文件路径: ${logsPath}`);
- resultStrArr.push(`[I] 标注文件数量: ${labelFiles.length}`);
- resultStrArr.push(`[I] 图片文件数量: ${imagesFiles.length}`);
- resultStrArr.push(`[I] 图片后缀: ${imageExts}`);
- // 判断labels中是否包含 classes.txt. 如果有则 将文件 复制至 transferPath 目录
- if (fs.existsSync(path.join(labelsPath, 'classes.txt'))) {
- await cpFile(path.join(labelsPath, 'classes.txt'), path.join(transferPath, 'classes.txt'))
- }
- // 将 resultArr 的内容, 转移至 logStrArr
- resultStrArr.forEach(item => {
- logStrArr.push(item);
- });
- for (const [i, labelFile] of labelFiles.entries()) {
- let filePath = path.join(labelsPath, labelFile);
- // 获取对应的图片路径, 与 label 文件的 名称相同
- let labelName = labelFile.replace('.txt', '');
- // 读取文件内容
- let fileData = await readFile(filePath);
- // 解析文件内容
- let result = parseYolo(fileData);
- // 图片后缀可能为 jpg, png 等
- let imagePath = '';
- let imageName = labelName;
- for (let j = 0; j < imageExts.length; j++) {
- imagePath = path.join(imagesPath, `${labelName}.${imageExts[0]}`);
- imageName = `${labelName}.${imageExts[0]}`;
- if (!fs.existsSync(imagePath)) {
- imagePath = '';
- continue;
- }
- }
- // 判断图片是否存在
- if (!imagePath) {
- logStrArr.push(`[E] 图片不存在 ${filePath} ${fileData} 可能是图片后缀异常 [ ${imageExts.join(', ')} ]`);
- continue;
- }
- readline.cursorTo(process.stdout, 0);
- readline.clearScreenDown(process.stdout);
- process.stdout.write(`${i} / ${labelFiles.length} | 解析文件: ${filePath} 中\n`);
- // console.log(result);
- // 空数据统计
- if (result.length === 0) {
- countMap['空'] = (countMap['空'] || 0) + 1;
- // 复制图片文件到 emptyPath
- let emptyFilePath = path.join(emptyPath, `${labelName}.${imageExts[0]}`);
- logStrArr.push(`[E] ${labelName}内容为空 空文件 ${emptyFilePath}`);
- console.log(`空文件: ${emptyFilePath}`)
- cpFile(imagePath, emptyFilePath).then(_=>_)
- continue;
- }
- // 获取文件标注类型
- let fileType = {};
- for (let j = 0; j < result.length; j++) {
- // let type = item.type;
- let typeName = dimensionType[result[j].type];
- if (!typeName) {
- console.log('未知类型')
- logStrArr.push(`[E] 未知类型 ${filePath} ${fileData}`);
- continue;
- }
- if (fileType[typeName]) {
- fileType[typeName]++;
- } else {
- fileType[typeName] = 1;
- }
- }
- console.log(fileType)
- if (Object.keys(fileType).length === 0) {
- countMap['异常标注'] = (countMap['异常标注'] || 0) + 1;
- continue;
- }
- // 获取标注的所有类型, 将图片进行区分 类型1:数量 类型2:数量 类型1-类型2:数量
- let fileTypeKeys = Object.keys(fileType);
- // 排序
- fileTypeKeys.sort();
- // 数据转换
- fileTypeKeys = fileTypeKeys.join('-');
- // 转换类型
- if (countMap[fileTypeKeys]) {
- countMap[fileTypeKeys]++;
- } else {
- countMap[fileTypeKeys] = 1;
- }
- // 尝试复制文件
- let transferFilePath = path.join(transferPath, fileTypeKeys);
- let transferImagePath = path.join(transferFilePath, `images/${imageName}` );
- let transferLabelPath = path.join(transferFilePath, `labels/${labelFile}`);
- // 文件转移
- // 同步复制文件
- cpFile(filePath, transferLabelPath).then(_=>_)
- cpFile(imagePath, transferImagePath).then(_=>_)
- logStrArr.push(`${fileTypeKeys}: ${countMap[fileTypeKeys]} ${filePath} ${imagePath} ===> ${transferLabelPath} ${transferImagePath} `);
- console.log(logStrArr[logStrArr.length - 1]);
- }
- // 输出结果
- console.log(countMap);
- // 将分析结果写入文件
- logStrArr.push(`\r\n${JSON.stringify(countMap, null, 4)}`);
- resultStrArr.push(`${JSON.stringify(countMap, null, 4)}`);
- // 保存结果
- let logStr = logStrArr.join('\r\n');
- let resultStr = resultStrArr.join('\r\n');
- await writeFile(path.join(logsPath, 'result.txt'), resultStr);
- await writeFile(path.join(logsPath, 'log.txt'), logStr);
- console.log('end');
- }
- // 将新处理的labels 重新写回至原目录
- function reCopyToRaw(){
- let labelsPath = `E:\\图库\\ai\\epower\\newLabels`
- // 数据转移目录
- let transferPath = `E:\\图库\\ai\\epower\\parse`
- // "0": "吊车",
- // "1": "塔吊",
- // "2": "烟火",
- // "3": "施工机械",
- // "4": "导线异物",
- // 吊车-导线异物
- // 施工机械-导线异物
- // 塔吊-导线异物
- // 塔吊-施工机械
- let fileTypeKeys = [1, 3]
- fileTypeKeys = fileTypeKeys.map(item => dimensionType[item] ? dimensionType[item] : '未知类型');
- let targetType = fileTypeKeys.join('-');
- let targetPath = path.join(transferPath, targetType);
- let targetLabelPath = path.join(targetPath, 'labels');
- let logStrArr = [];
- // 遍历目录
- let labelFiles = fs.readdirSync(targetLabelPath);
- for (let i = 0; i < labelFiles.length; i++) {
- let labelFilePath = path.join(targetLabelPath, labelFiles[i]);
- // 拷贝文件
- cpFile(labelFilePath, path.join(labelsPath, labelFiles[i])).then(_=>_)
- }
- console.log(targetPath);
- }
- // 复制分类好的label文件至指定目录
- async function copyLabel() {
- // let allLabelsPath = `E:\\图库\\ai\\epower_v2\\parse`
- let allLabelsPath = `E:\\图库\\ai\\epower\\parse`
- // let transferPath = `E:\\图库\\ai\\epower_v2\\tmpLabels_2`
- let transferPath = `E:\\图库\\ai\\epower\\tmpLabels_2`
- // 获取子目录
- let subDirs = fs.readdirSync(allLabelsPath);
- for (let i = 0; i < subDirs.length; i++) {
- if (subDirs[i] == 'classes.txt'){
- console.log(`skip classes.txt`)
- continue;
- }
- let subDir = path.join(allLabelsPath, subDirs[i], 'labels');
- console.log(subDir)
- // 获取子目录下的labels
- let labelFiles = fs.readdirSync(subDir);
- for (let j = 0; j < labelFiles.length; j++) {
- // 获取文件路径
- let labelFilePath = path.join(subDir, labelFiles[j]);
- // 复制文件 至新目录 新目录: 新目录/子目录名称
- let newDir = path.join(transferPath, subDirs[i]);
- let newFilePath = path.join(newDir, labelFiles[j]);
- // 复制文件
- await cpFile(labelFilePath, newFilePath)
- }
- }
- }
- // 将转移出来的label文件合并至一个文件夹内
- async function copyLabelToAll() {
- let allLabelsPath = `E:\\图库\\ai\\epower\\tmpLabels_2`
- let baseImagesPath = `E:\\图库\\ai\\epower\\images`
- let transferPath = `E:\\图库\\ai\\epower\\all_tmpLabels_2`
- let trabsferImaegsPath = `E:\\图库\\ai\\epower\\all_tmpImages_2`
- let imageExts = ['jpg']
- // 获取子目录
- let subDirs = fs.readdirSync(allLabelsPath);
- for (let i = 0; i < subDirs.length; i++) {
- if (subDirs[i] == 'classes.txt'){
- console.log(`skip classes.txt`)
- continue;
- }
- let subDir = path.join(allLabelsPath, subDirs[i]);
- console.log(subDir)
- // 获取子目录下的labels
- let labelFiles = fs.readdirSync(subDir);
- for (let j = 0; j < labelFiles.length; j++) {
- if (labelFiles[j] == 'classes.txt'){
- console.log(`skip classes.txt`)
- continue;
- }
- let labelName = labelFiles[j].replace('.txt', '');
- // 获取文件路径
- let labelFilePath = path.join(subDir, labelFiles[j]);
- // 复制文件 至新目录 新目录: 新目录/子目录名称
- let newFilePath = path.join(transferPath, labelFiles[j]);
- // 复制文件
- cpFile(labelFilePath, newFilePath)
- // 复制图片
- // 图片后缀可能为 jpg, png 等
- let imagePath = '';
- let imageName = `${labelName}.jpg`;
- imagePath = path.join(baseImagesPath, imageName);
- // 复制图片
- let newImagePath = path.join(trabsferImaegsPath, imageName);
- console.log(`移动图片 ${imagePath} => ${newImagePath}`)
- // 判断图片是否存在
- if (!imagePath) {
- console.log(`[E] 图片不存在 ${labelFilePath} 可能是图片后缀异常 [ ${imageExts.join(', ')} ]`);
- continue;
- }
- cpFile(imagePath, newImagePath)
- }
- }
- }
- function replaceNewlines(input) {
- return input.replace(/(\r\n|\r|\n)/g, '\r\n');
- }
- // 将label 文件中的换行 从 \n 替换为 \r\n
- function relineLabel(){
- let labelsPath = `E:\\图库\\ai\\epower\\all_tmpLabels_2`
- // 遍历目录
- let labelFiles = fs.readdirSync(labelsPath);
- for (let i = 0; i < labelFiles.length; i++) {
- let labelFilePath = path.join(labelsPath, labelFiles[i]);
- if (!fs.statSync(labelFilePath).isFile()){
- continue;
- }
- console.log(`start reline ${labelFilePath}`);
- let fileContent = fs.readFileSync(labelFilePath, 'utf-8');
- fileContent = replaceNewlines(fileContent);
- fs.writeFileSync(labelFilePath, fileContent)
- }
- }
- main();
- // reCopyToRaw();
- // copyLabel();
- // copyLabelToAll()
- // relineLabel();
|