检验转换后的paddleseg标注图像和标注之间的对应关系（假彩色）

对比标注和标注图之间是否一致（每一张图可视化）

import os
import numpy as np
from PIL import Image, ImageDraw, ImageFont
import cv2

# ====== 输入路径（根据输出路径设置）======
dataset_dir = './data/test1'
image_dir = os.path.join(dataset_dir, 'images')
label_dir = os.path.join(dataset_dir, 'annotations')
class_file = os.path.join(dataset_dir, 'class_names.txt')
output_vis_dir = os.path.join(dataset_dir, 'vis_check')
os.makedirs(output_vis_dir, exist_ok=True)

# ====== 加载类别映射 ======
with open(class_file, 'r', encoding='utf-8') as f:
    class_names = [line.strip() for line in f.readlines()]


# ====== 创建颜色映射 ======
def get_color_map_list(num_classes):
    num_classes += 1
    color_map = num_classes * [0, 0, 0]
    for i in range(0, num_classes):
        j = 0
        lab = i
        while lab:
            color_map[i * 3] |= (((lab >> 0) & 1) << (7 - j))
            color_map[i * 3 + 1] |= (((lab >> 1) & 1) << (7 - j))
            color_map[i * 3 + 2] |= (((lab >> 2) & 1) << (7 - j))
            j += 1
            lab >>= 3
    return np.array(color_map[3:]).reshape((-1, 3)).astype(np.uint8)


palette = get_color_map_list(256)

# ====== 可视化并检查每张图 ======
for fname in os.listdir(image_dir):
    name, ext = os.path.splitext(fname)
    image_path = os.path.join(image_dir, fname)
    label_path = os.path.join(label_dir, name + '.png')

    if not os.path.exists(label_path):
        print(f"⚠️ 缺失标注图：{label_path}")
        continue

    # 加载图像和标注
    img = Image.open(image_path).convert("RGB")
    label = Image.open(label_path).convert("P")
    label_np = np.array(label)
    label_rgb = palette[label_np]

    # 创建透明融合图
    blended = Image.blend(img, Image.fromarray(label_rgb), alpha=0.8)
    draw = ImageDraw.Draw(blended)

    # 使用字体
    try:
        font = ImageFont.truetype("arial.ttf", 14)
    except:
        font = ImageFont.load_default()

    # 标注每个类别的重心
    used_class_ids = []
    for class_id in np.unique(label_np):
        if class_id == 0:
            continue  # 跳过背景
        mask = (label_np == class_id).astype(np.uint8)
        if mask.sum() == 0:
            continue
        M = cv2.moments(mask)
        if M['m00'] == 0:
            continue
        cx = int(M['m10'] / M['m00'])
        cy = int(M['m01'] / M['m00'])
        class_name = class_names[class_id]
        draw.text((cx, cy), class_name, fill='white', font=font)
        used_class_ids.append(class_id)

    # 添加图例（legend）
    legend_height = 20 * len(used_class_ids)
    legend_width = 200
    legend = Image.new('RGB', (legend_width, legend_height), color=(255, 255, 255))
    legend_draw = ImageDraw.Draw(legend)
    for idx, class_id in enumerate(used_class_ids):
        color = tuple(palette[class_id])
        class_name = class_names[class_id]
        y = idx * 20
        legend_draw.rectangle([0, y, 20, y + 15], fill=color)
        legend_draw.text((25, y), class_name, fill='black', font=font)

    # 合并图像和图例
    combined = Image.new('RGB', (blended.width + legend_width, max(blended.height, legend_height)), (255, 255, 255))
    combined.paste(blended, (0, 0))
    combined.paste(legend, (blended.width, 0))

    # 保存可视化图
    out_path = os.path.join(output_vis_dir, name + '_vis.png')
    combined.save(out_path)
    print(f"✅ 可视化完成: {out_path}")

print("\n🎉 检查与可视化全部完成！输出目录：", output_vis_dir)