diff --git a/src/V2X_edge_intelligence/src/README.md b/src/V2X_edge_intelligence/src/README.md new file mode 100644 index 0000000000..1778860e16 --- /dev/null +++ b/src/V2X_edge_intelligence/src/README.md @@ -0,0 +1,12 @@ +# 路侧感知数据集预处理模块(Carla适配) +## 模块说明 +本模块针对Carla模拟器生成的车路协同路侧感知数据,实现轻量化预处理(数据生成、增强、划分),全程无第三方依赖,直接运行出结果。 + +## 运行环境 +- Python 3.6+(无需任何第三方库,仅用内置模块) +- 操作系统:Windows/Linux/macOS均可 + +## 运行步骤 +1. 进入模块目录: + ```bash + cd C:\Users\21015\PycharmProjects\nn2\V2X_edge_intelligence\src \ No newline at end of file diff --git a/src/V2X_edge_intelligence/src/main.py b/src/V2X_edge_intelligence/src/main.py new file mode 100644 index 0000000000..8143a8ec03 --- /dev/null +++ b/src/V2X_edge_intelligence/src/main.py @@ -0,0 +1,92 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +""" +路侧感知数据集预处理(Carla适配) +零第三方依赖(仅Python内置库),完全无报错! +运行方式:python main.py +""" +import json +import os +import random + + +# ===================== 第一步:生成模拟Carla数据(纯文本,不用图片) ===================== +def generate_demo_data(): + """生成模拟Carla标注数据(纯文本,无需图片/OpenCV,零报错)""" + os.makedirs("demo_carla_data", exist_ok=True) + # 生成Carla场景标注(纯JSON文本,模拟感知数据) + anno_data = { + "carla_scenes": [ + {"scene_id": 1001, "frame_id": 0, + "obstacles": [{"type": "car", "bbox": [100, 100, 200, 200], "distance": 8.5}]}, + {"scene_id": 1002, "frame_id": 1, + "obstacles": [{"type": "person", "bbox": [150, 150, 250, 250], "distance": 5.2}]} + ] + } + with open("demo_carla_data/carla_anno.json", "w", encoding="utf-8") as f: + json.dump(anno_data, f, indent=2) + print("✅ 模拟Carla标注数据生成完成 → demo_carla_data/carla_anno.json") + + +# ===================== 第二步:数据增强(文本层面模拟,无需图片) ===================== +def simple_augment(): + """模拟数据增强(文本层面扩充,比如添加噪声、复制数据)""" + with open("demo_carla_data/carla_anno.json", "r", encoding="utf-8") as f: + data = json.load(f) + + # 模拟增强:为每个场景添加随机噪声(模拟图像增强) + augmented_data = [] + for scene in data["carla_scenes"]: + # 复制场景并添加噪声 + aug_scene = scene.copy() + aug_scene["aug_type"] = "random_brightness" # 模拟亮度增强 + # 给障碍物距离加随机噪声 + for obs in aug_scene["obstacles"]: + obs["distance"] = round(obs["distance"] + random.uniform(-0.5, 0.5), 2) + augmented_data.append(aug_scene) + + # 保存增强后数据 + with open("demo_carla_data/carla_anno_augmented.json", "w", encoding="utf-8") as f: + json.dump({"carla_scenes_augmented": augmented_data}, f, indent=2) + print("✅ 数据增强完成 → demo_carla_data/carla_anno_augmented.json") + + +# ===================== 第三步:数据集划分(纯文本,内置库实现) ===================== +def split_dataset(): + """划分数据集(8:1:1,纯文本处理)""" + with open("demo_carla_data/carla_anno.json", "r", encoding="utf-8") as f: + data = json.load(f) + + scenes = data["carla_scenes"] + random.shuffle(scenes) # 随机打乱 + total = len(scenes) + train_size = int(total * 0.8) + val_size = int((total - train_size) / 2) + + # 划分数据 + train_data = scenes[:train_size] + val_data = scenes[train_size:train_size + val_size] + test_data = scenes[train_size + val_size:] + + # 保存划分结果 + split_result = { + "train_scenes": train_data, + "val_scenes": val_data, + "test_scenes": test_data, + "split_ratio": "train:80% | val:10% | test:10%" + } + with open("demo_carla_data/carla_split_result.json", "w", encoding="utf-8") as f: + json.dump(split_result, f, indent=2) + print("✅ 数据集划分完成 → demo_carla_data/carla_split_result.json") + print(f" 划分结果:训练集{len(train_data)}条 | 验证集{len(val_data)}条 | 测试集{len(test_data)}条") + + +# ===================== 主函数:一键运行 ===================== +if __name__ == "__main__": + print("===== 路侧感知数据集预处理(Carla适配) =====\n") + generate_demo_data() + simple_augment() + split_dataset() + print("\n🎉 所有步骤运行完成!生成文件列表:") + for file in os.listdir("demo_carla_data"): + print(f" - demo_carla_data/{file}") \ No newline at end of file