当前位置：首页 > news >正文

coco128数据集格式

news 来源：原创 2025/4/21 16:43:29

一、数据集内容

每一行表示一个目标的信息：5个字段（class_id, x_center, y_center, width, height）

二、可视化标注框与类别

import os
import cv2
import matplotlib.pyplot as plt# 设置路径
data_dir = "F:/datasets/coco128"  # 替换为你的实际路径
image_folder = os.path.join(data_dir, "images/train2017")  # 选择train/val/test
label_folder = os.path.join(data_dir, "labels/train2017")  # 对应标签文件夹# 手动指定图像文件名（例如："0000001.jpg"）
target_image_name = "000000000025.jpg"  # 修改为目标图像名
image_path = os.path.join(image_folder, target_image_name)
label_path = os.path.join(label_folder, os.path.splitext(target_image_name)[0] + '.txt')# 检查文件是否存在
if not os.path.exists(image_path):print(f"错误：图像文件 {image_path} 不存在！")exit()
if not os.path.exists(label_path):print(f"错误：标签文件 {label_path} 不存在！")exit()# 读取图像
image = cv2.imread(image_path)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)  # 转换为RGB格式
img_h, img_w = image.shape[:2]  # 获取图像高宽# 读取并绘制YOLO标签
with open(label_path, 'r') as f:for line in f.readlines():parts = line.strip().split()if len(parts) < 5:continue  # 跳过格式错误的行class_id, x_center, y_center, width, height = map(float, parts[:5])# 转换为绝对坐标（像素值）x_center *= img_wy_center *= img_hwidth *= img_wheight *= img_hx_min = int(x_center - width / 2)y_min = int(y_center - height / 2)x_max = int(x_center + width / 2)y_max = int(y_center + height / 2)# 绘制边界框（绿色框，线宽2像素）cv2.rectangle(image, (x_min, y_min), (x_max, y_max), (0, 255, 0), 2)# 仅显示类别数值（class_id）cv2.putText(image, f"{int(class_id)}",(x_min, y_min - 10),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)# 显示结果
plt.figure(figsize=(12, 8))
plt.imshow(image)
plt.axis('off')
plt.title(f"Image: {target_image_name} (Class IDs only)")
plt.show()