Yolo11 基于DroneVehicle数据集的无人机视角下车辆目标检测

Yolo11 基于DroneVehicle数据集的无人机视角下车辆目标检测 | 极客日志

import numpy as np import cv2 import os from tqdm import tqdm def create_file(output_dir_vi, output_dir_ir): if not os.path.exists(output_dir_vi): os.makedirs(output_dir_vi) if not os.path.exists(output_dir_ir): os.makedirs(output_dir_ir) print(f'Created folder:({output_dir_vi}); ({output_dir_ir})') def update(input_img_path, output_img_path): image = cv2.imread(input_img_path) cropped = image[100:612, 100:740] # 裁剪坐标为[y0:y1, x0:x1] cv2.imwrite(output_img_path, cropped) dataset_dir_vi = r'valimg' # 处理前可见光图片目录 output_dir_vi = r'valimg2' # 处理后可见光图片目录 dataset_dir_ir = r'valimgr' # 处理前红外光图片目录 output_dir_ir = r'valimgr2' # 处理后红外光图片目录 # 检查文件夹是否存在，如果不存在则创建 create_file(output_dir_vi, output_dir_ir) # 获得需要转化的图片路径并生成目标路径 image_filenames_vi = [(os.path.join(dataset_dir_vi, x), os.path.join(output_dir_vi, x)) for x in os.listdir(dataset_dir_vi)] image_filenames_ir = [(os.path.join(dataset_dir_ir, x), os.path.join(output_dir_ir, x)) for x in os.listdir(dataset_dir_ir)] # 转化所有图片 print('Start transforming vision images...') for path in tqdm(image_filenames_vi): update(path[0], path[1]) print('Start transforming infrared images...') for path in tqdm(image_filenames_ir): update(path[0], path[1])

import xml.etree.ElementTree as ET import shutil import os import imagesize # 定义识别目标或类集合 object = 'datasets' # 根据自定义的数据集名称 if os.path.exists("./%s/labels/"%object): # 如果文件存在 shutil.rmtree("./%s/labels/"%object) os.makedirs("./%s/labels/"%object) else: os.makedirs("./%s/labels/"%object) sets = ['train', 'val'] # 修改类别(自定义) classes =["car", "truck", "bus", "van", "freight_car"] def convert(size, box): # 坐标信息归一化至0-1 dw = 1. / size[0] dh = 1. / size[1] x = (box[0] + box[1]) / 2.0 y = (box[2] + box[3]) / 2.0 w = box[1] - box[0] h = box[3] - box[2] x = x * dw w = w * dw y = y * dh h = h * dh return (x, y, w, h) def convert_annotation(image_id): in_file = open('./%s/xml/%s.xml' % (object,image_id)) # xml文件 out_file = open('./%s/labels/%s.txt' % (object,image_id), 'w') # txt文件 image_file = open('./%s/images/%s.jpg' % (object,image_id)) # pic文件 print("in_file,",in_file) tree = ET.parse(in_file) # f = open(in_file.name,encoding="utf-8") # tree = ET.parse(f) root = tree.getroot() size = root.find('size') # 这里的width 和 height 在Autolabelimg下自动标注可能会被修改，需替换成图片的真实宽高 # w = int(size.find('width').text) # h = int(size.find('height').text) w, h = imagesize.get(image_file.name) for obj in root.iter('object'): difficult = obj.find('difficult').text cls = obj.find('name').text if cls not in classes or int(difficult) == 1: continue cls_id = classes.index(cls) # 类别序号 xmlbox = obj.find('bndbox') xmin = float(xmlbox.find('xmin').text) xmin = xmin if xmin >= 0 else 0.0 # 左上角x坐标如果小于0都化成0 xmax = float(xmlbox.find('xmax').text) xmax = xmax if xmax <= w else float(w) # 右下角x坐标如果大于图片宽度了都为图片宽度值 ymin = float(xmlbox.find('ymin').text) ymin = ymin if ymin >= 0 else 0.0 # 左上角y坐标如果小于0都化成0 ymax = float(xmlbox.find('ymax').text) ymax = ymax if ymax <= h else float(h) # 右下角y坐标如果大于图片高度了都为图片高度值 b = (xmin,xmax ,ymin ,ymax) bb = convert((w, h), b) # 归一化 out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n') for image_set in sets: if not os.path.exists('./%s/labels/'%object): os.makedirs('./%s/labels/'%object) image_ids = open('./%s/ImageSets/%s.txt' % (object,image_set)).read().strip().split() list_file = open('./%s/%s.txt' % (object,image_set), 'w') for image_id in image_ids: list_file.write('./images/%s.jpg\n' % (image_id)) # 要注意图片的后缀名是什么 convert_annotation(image_id) list_file.close()