使用Yolov8 训练交通标志数据集：TT100K数据集划分

---

一、数据集下载

1. 官方网址：TT100K
   

2. 数据集对比
   
   源码如下：

   def classes(filedir):with open(filedir) as f:classes = json.load(f)classes_list = classes['types']return len(classes_list)c_2016 = classes("./annotations.json")
   c_2021 = classes("./annotations_all.json")
   print(f'TT100K-2016 交通标志的类别：{c_2016}')
   print(f'TT100K-2021 交通标志的类别：{c_2021}')
   print(f'差值：{c_2021-c_2016}')
   

3. 查看类别数目

   # 读TT100K原始数据集标注文件
   with open(annotations) as p:# 读取 json 文件file = json.load(p)# 获取图字典img_dict = file['imgs']types = file['types']classes_list = dict()for i in types:# i 为所有的类别： 如：‘pl80’, 'w9', 'p6'# 将每一个类别作为字典中的键，值为: 空列表 []classes_list[i] = 0for img_id in img_dict:img = images_dict[img_id]obj_list = img['objects']# 统计实际有多少种类别for obj in obj_list:classes_list[obj['category']] += 1# 大于0的有201个d = []for keym, val in classes_list.items():if val > 0:d.append(val)print(len(d))
   

二、划分数据集

数据集划分有四个步骤，依次按照一下步骤执行即可：

1. 获取每个类别有多少图片的 classes_statistics.json 文件。此过程会产生一个 json 文件，保存了分类类别，每个类别的详细信息，以及所有图片信息。
   
   
   源码如下：

   def classes_statistics():"""类别统计函数，统计每一个类别下有多少张图片保存在 classes_statistics.json 文件中:return:"""# 读TT100K原始数据集标注文件with open(annotations) as origin_json:# 读取 json 文件origin_dict = json.load(origin_json)# 获取类别列表classes_list = origin_dict['types']# 获取图字典images_dict = origin_dict['imgs']# 建立统计每个类别包含的图片的字典sta = dict()for i in classes_list:# i 为所有的类别： 如：‘pl80’, 'w9', 'p6'# 将每一个类别作为字典中的键，值为: 空列表 []sta[i] = []# 遍历TT100K的imgsfor image_id in images_dict:image_element = images_dict[image_id]image_path = image_element['path']# 添加图像的信息到dataset中image_path = image_path.split('/')[-1]  # 如：‘62627.jpg’obj_list = image_element['objects']# 遍历每张图片的标注信息，将每张图片的路径添加到对应类别的字典中去for anno_dic in obj_list:label_key = anno_dic['category']# 防止一个图片多次加入一个标签类别if image_path not in sta[label_key]:sta[label_key].append(image_path)# 数据清洗--只保留包含图片数超过100的类别# 清洗之后还剩下：45种类别result = {k: v for k, v in sta.items() if len(v) > 100}# 将所有标注写入文件 此时写入超过100张图片的with open('./all_classes.json', 'w', encoding="utf-8") as f:json.dump(result, f, ensure_ascii=False, indent=2)# 记录所有保留的图片saved_images = []for i in result:saved_images.extend(result[i])# 第二次数据清洗。同一张图片有多个类别标注，所以要去重old_saved_images = len(saved_images)saved_images = list(set(saved_images))new_saved_images = len(saved_images)# print(f"total types is: {len(result)}")type_list = list(result.keys())# 字典，三个键：类型(列表)，详细信息(字典)，所有类型的图片数据(列表)result2 = {"type": type_list, "details": result, "images": saved_images}# 保存结果json_name = classes_statistics_namewith open(json_name, 'w', encoding="utf-8") as f:json.dump(result2, f, ensure_ascii=False, indent=2)print(f"类别统计文件已保存：{json_name}")return type_list, result, saved_images, result2, old_saved_images, new_saved_images
   

   结果如下：
   

2. 将 TT100K 原始数据集转换为 COCO 数据集格式，并根据图片中各类别的数量比例将数据划分为训练集、验证集和测试集，以便后续的模型训练和评估。此步骤生成，train.json， test.json， val.json 文件，划分比例为7：2：1。
   源码如下：

   def original_datasets2object_datasets_re():'''重新划分数据集将 TT100K 原始数据集转换为 COCO 数据集格式，并根据图片中各类别的数量比例将数据划分为训练集、验证集和测试集，以便后续的模型训练和评估。:return:'''# 读TT100K原始数据集标注文件with open(annotations) as origin_json:origin_dict = json.load(origin_json)with open(classes_statistics_name) as select_json:select_dict = json.load(select_json)classes = select_dict['type']train_dataset = {'info': {}, 'licenses': [], 'categories': [], 'images': [], 'annotations': []}val_dataset = {'info': {}, 'licenses': [], 'categories': [], 'images': [], 'annotations': []}test_dataset = {'info': {}, 'licenses': [], 'categories': [], 'images': [], 'annotations': []}label = {}  # 记录每个标志类别的idcount = {}  # 记录每个类别的图片数owntype_sum = {}info = {"year": 2021,  # 年份"version": '1.0',  # 版本"description": "TT100k_to_coco",  # 数据集描述"contributor": "Tecent&Tsinghua",  # 提供者"url": 'https://cg.cs.tsinghua.edu.cn/traffic-sign/',  # 下载地址"date_created": 2021 - 1 - 15}licenses = {"id": 1,"name": "null","url": "null",}train_dataset['info'] = infoval_dataset['info'] = infotest_dataset['info'] = infotrain_dataset['licenses'] = licensesval_dataset['licenses'] = licensestest_dataset['licenses'] = licenses# 建立类别和id的关系for i, cls in enumerate(classes):train_dataset['categories'].append({'id': i, 'name': cls, 'supercategory': 'traffic_sign'})val_dataset['categories'].append({'id': i, 'name': cls, 'supercategory': 'traffic_sign'})test_dataset['categories'].append({'id': i, 'name': cls, 'supercategory': 'traffic_sign'})label[cls] = icount[cls] = 0owntype_sum[cls] = 0images_dic = origin_dict['imgs']obj_id = 1# 计算出每个类别共‘包含’的图片数for image_id in images_dic:image_element = images_dic[image_id]image_path = image_element['path']image_name = image_path.split('/')[-1]# 在所选的类别图片中if image_name not in select_dict['images']:continue# 处理TT100K中的标注信息obj_list = image_element['objects']# 记录图片中包含最多的实例所属的typeincludes_type = {}for anno_dic in obj_list:if anno_dic["category"] not in select_dict["type"]:continue# print(anno_dic["category"])if anno_dic["category"] in includes_type:includes_type[anno_dic["category"]] += 1else:includes_type[anno_dic["category"]] = 1# print(includes_type)own_type = max(includes_type, key=includes_type.get)owntype_sum[own_type] += 1# TT100K的annotation_all转换成coco的格式for image_id in images_dic:image_element = images_dic[image_id]image_path = image_element['path']image_name = image_path.split('/')[-1]# 在所选的类别图片中if image_name not in select_dict['images']:continueprint(f"dealing with {image_path} ")# 处理TT100K中的标注信息obj_list = image_element['objects']# 记录图片中包含最多的实例所属的typeincludes_type = {}for anno_dic in obj_list:if anno_dic["category"] not in select_dict["type"]:continue# print(anno_dic["category"])if anno_dic["category"] in includes_type:includes_type[anno_dic["category"]] += 1else:includes_type[anno_dic["category"]] = 1# print(includes_type)own_type = max(includes_type, key=includes_type.get)count[own_type] += 1num_rate = count[own_type] / owntype_sum[own_type]# 划分数据集 7：2：1，train_set,val_set,test_set。if num_rate < 0.7:dataset = train_datasetelif num_rate < 0.9:dataset = val_datasetelse:print("dataset=test_dataset")dataset = test_datasetfor anno_dic in obj_list:if anno_dic["category"] not in select_dict["type"]:continuex = anno_dic['bbox']['xmin']y = anno_dic['bbox']['ymin']width = anno_dic['bbox']['xmax'] - anno_dic['bbox']['xmin']height = anno_dic['bbox']['ymax'] - anno_dic['bbox']['ymin']label_key = anno_dic['category']dataset['annotations'].append({'area': width * height,'bbox': [x, y, width, height],'category_id': label[label_key],'id': obj_id,'image_id': image_id,'iscrowd': 0,'segmentation': [[x, y, x + width, y, x + width, y + height, x, y + height]]})# 每个标注的对象id唯一obj_id += 1im = cv2.imread(os.path.join(datasets_dir, image_path))H, W, _ = im.shape# 添加图像的信息到dataset中dataset['images'].append({'file_name': image_name,'id': image_id,'width': W,'height': H})# 保存结果for phase in ['train', 'val', 'test']:# 创建保存位置的文件夹os.makedirs(RETT100K_dir+'/labels', exist_ok=True)json_name = RETT100K_dir+f'/labels/{phase}.json'with open(json_name, 'w', encoding="utf-8") as f:if phase == 'train':json.dump(train_dataset, f, ensure_ascii=False, indent=2)if phase == 'val':json.dump(val_dataset, f, ensure_ascii=False, indent=2)if phase == 'test':json.dump(test_dataset, f, ensure_ascii=False, indent=2)
   

   执行结果：
   
   

3. 写入归一化的标注文件， 生成 yolo 格式的标注信息。获取标注信息，并划分 yolo 格式的标注信息 *.txt 文件夹
   源码如下：

   def coco_json2yolo_txt(class_json):# 将标注位置坐标做归一化处理def convert(size, box):dw = 1. / (size[0])dh = 1. / (size[1])x = box[0] + box[2] / 2.0y = box[1] + box[3] / 2.0w = box[2]h = box[3]x = round(x * dw, 6)w = round(w * dw, 6)y = round(y * dh, 6)h = round(h * dh, 6)return (x, y, w, h)json_file = os.path.join(RETT100K_dir+f'/labels/{class_json}.json')ana_txt_save_path = os.path.join(RETT100K_dir+'labels', class_json)  # 保存的路径data = json.load(open(json_file, 'r'))if not os.path.exists(ana_txt_save_path):os.makedirs(ana_txt_save_path)id_map = {}  # coco数据集的id不连续！重新映射一下再输出！with open(os.path.join(ana_txt_save_path, 'classes.txt'), 'w') as f:# 写入classes.txtfor i, category in enumerate(data['categories']):f.write(f"{category['name']}\n")id_map[category['id']] = i# 写入图像相对路径的文件位置list_file = open(os.path.join(ana_txt_save_path, '%s.txt' % class_json.format()), 'w')for img in tqdm(data['images']):filename = img["file_name"]img_width = img["width"]img_height = img["height"]img_id = img["id"]head, tail = os.path.splitext(filename)ana_txt_name = head + ".txt"  # 对应的txt名字，与jpg一致f_txt = open(os.path.join(ana_txt_save_path, ana_txt_name), 'w')for ann in data['annotations']:if ann['image_id'] == img_id:box = convert((img_width, img_height), ann["bbox"])f_txt.write("%s %s %s %s %s\n" % (id_map[ann["category_id"]], box[0], box[1], box[2], box[3]))f_txt.close()# 将图片的相对路径写入文件list_file.write('/%s/%s.jpg\n' % (class_json.format(), head))list_file.close()
   

   # 3. 写入归一化的标注文件， 生成 yolo 格式的标注信息
   phase=['train', 'val', 'test']
   for phase in phase:coco_json2yolo_txt(phase)
   

   执行结果：
   
   

4. 拷贝图片，数据集划分完成。
   源码如下：

   def divide_TrainValTest(source, target, pre_datasets):'''创建文件路径并复制数据集:param source: 源文件位置:param target: 目标文件位置:param pre_datasets: 预处理数据集路径'''# 创建输出目录for i in ['train', 'val', 'test']:os.makedirs(f"{target}/{i}", exist_ok=True)# 获取当前处理的数据集类型dataset_type = Noneif 'train' in source:dataset_type = 'train'elif 'val' in source:dataset_type = 'val'elif 'test' in source:dataset_type = 'test'# 收集所有需要处理的文件file_list = []for root, dirs, files in os.walk(source):for file in files:file_name = os.path.splitext(file)[0]# 排除无关项if file_name in ['classes', 'test', 'train', 'val', '10106-checkpoint']:continuefile_list.append(file_name)for file_name in tqdm(file_list, desc=f"处理{dataset_type}数据集", unit="文件"):image_path = f"{file_name}.jpg"source_path = os.path.join(pre_datasets, image_path)target_path = os.path.join(target, dataset_type, image_path)# 复制文件并处理异常try:shutil.copyfile(source_path, target_path)except Exception as e:print(f"警告: 无法复制文件 {image_path}: {e}")print(f"{dataset_type}数据集处理完成！")
   

   执行结果：
   

三、目录放置

1. 原数据集的目录结构：保持下载时目录不变即可。

   E:\机器学习\TT00K\tt100k_2021>tree
   E:.
   ├─train
   ├─test
   └─other
   

2. 处理后的数据集目录结：

   E:\机器学习\TT00K\RETT100K_2021>tree
   E:.
   ├─images
   │  ├─test
   │  ├─train
   │  └─val
   └─labels├─test├─train└─val
   

3. 处理完之后就可以使用 yolo v8/v11 训练了。
   data.yaml 文件如下：

   path: ../datasets/RE_TT100K_2021/
   train: images/train
   val: images/val
   test: images/test# 分类总数
   nc: 45
   # 分类类别
   names: ['pl80', 'p6', 'p5', 'pm55', 'pl60', 'ip', 'p11', 'i2r', 'p23', 'pg', 'il80', 'ph4', 'i4', 'pl70', 'pne', 'ph4.5', 'p12', 'p3', 'pl5', 'w13', 'i4l', 'pl30', 'p10', 'pn', 'w55', 'p26', 'p13', 'pr40', 'pl20', 'pm30', 'pl40', 'i2', 'pl120', 'w32', 'ph5', 'il60', 'w57', 'pl100', 'w59', 'il100', 'p19', 'pm20', 'i5', 'p27', 'pl50']
   

   

---

END

---