无人机建筑垃圾建材目标检测数据集 航拍建筑垃圾检测数据集1111CDW建筑垃圾建材目标检测数据集数据集信息表项目详细说明数据集名称CDW建筑垃圾建材目标检测数据集图片总量2204张2.2k检测类别共12类metal金属、wood木材、empty空地、cover遮盖物、electronic-waste电子垃圾、gravel砂石、insulator-wool保温岩棉、mixed-waste混合废弃物、plasterboard石膏板、polystyrene聚苯乙烯泡沫、roofing-felt屋面油毡、wood-panel木饰面板模型基准精度mAP5098.5%精确率(Precision)92.0%召回率(Recall)97.8%数据格式支持标准YOLO、COCO、VOC等多种标注格式可按需导出适用场景建筑垃圾智能分拣、航拍建筑固废识别、工地物料分类、再生资源AI识别、环境监测、目标检测模型训练数据集配置文件cdw_materials.yamlpath:./cdw_materials_datasettrain:images/trainval:images/valtest:images/testnc:12names:0:metal1:wood2:empty3:cover4:electronic-waste5:gravel6:insulator-wool7:mixed-waste8:plasterboard9:polystyrene10:roofing-felt11:wood-panel环境依赖pipinstallultralytics torch opencv-python训练代码train_cdw.pyfromultralyticsimportYOLOdefmain():# 加载YOLO预训练模型兼容YOLOv8/YOLOv11modelYOLO(yolov8s.pt)train_resultsmodel.train(data./cdw_materials_dataset/cdw_materials.yaml,epochs100,imgsz640,batch16,device0,# 无GPU改为 devicecpuworkers4,patience15,projectruns/train,namecdw_material_detect,exist_okTrue,pretrainedTrue)print(训练完成权重路径runs/train/cdw_material_detect/weights)# 测试集评估metricsmodel.val(data./cdw_materials_dataset/cdw_materials.yaml,splittest)print(fmAP50:{metrics.box.map50:.4f})if__name____main__:main()推理测试代码predict_cdw.pyfromultralyticsimportYOLOimportcv2# 加载训练最优权重modelYOLO(runs/train/cdw_material_detect/weights/best.pt)if__name____main__:img_pathcdw_test.jpgresultsmodel.predict(sourceimg_path,conf0.25,iou0.45,saveTrue)result_imgresults[0].plot()cv2.imshow(CDW Materials Detection,result_img)cv2.waitKey(0)cv2.destroyAllWindows()核心标签#建筑垃圾检测#固废识别#建材分类#YOLO数据集#工地物料识别#航拍垃圾巡检
YOLO模型如何训练 无人机建筑垃圾建材目标检测数据集 航拍建筑垃圾检测数据集
发布时间：2026/7/28 11:48:37
无人机建筑垃圾建材目标检测数据集 航拍建筑垃圾检测数据集1111CDW建筑垃圾建材目标检测数据集数据集信息表项目详细说明数据集名称CDW建筑垃圾建材目标检测数据集图片总量2204张2.2k检测类别共12类metal金属、wood木材、empty空地、cover遮盖物、electronic-waste电子垃圾、gravel砂石、insulator-wool保温岩棉、mixed-waste混合废弃物、plasterboard石膏板、polystyrene聚苯乙烯泡沫、roofing-felt屋面油毡、wood-panel木饰面板模型基准精度mAP5098.5%精确率(Precision)92.0%召回率(Recall)97.8%数据格式支持标准YOLO、COCO、VOC等多种标注格式可按需导出适用场景建筑垃圾智能分拣、航拍建筑固废识别、工地物料分类、再生资源AI识别、环境监测、目标检测模型训练数据集配置文件cdw_materials.yamlpath:./cdw_materials_datasettrain:images/trainval:images/valtest:images/testnc:12names:0:metal1:wood2:empty3:cover4:electronic-waste5:gravel6:insulator-wool7:mixed-waste8:plasterboard9:polystyrene10:roofing-felt11:wood-panel环境依赖pipinstallultralytics torch opencv-python训练代码train_cdw.pyfromultralyticsimportYOLOdefmain():# 加载YOLO预训练模型兼容YOLOv8/YOLOv11modelYOLO(yolov8s.pt)train_resultsmodel.train(data./cdw_materials_dataset/cdw_materials.yaml,epochs100,imgsz640,batch16,device0,# 无GPU改为 devicecpuworkers4,patience15,projectruns/train,namecdw_material_detect,exist_okTrue,pretrainedTrue)print(训练完成权重路径runs/train/cdw_material_detect/weights)# 测试集评估metricsmodel.val(data./cdw_materials_dataset/cdw_materials.yaml,splittest)print(fmAP50:{metrics.box.map50:.4f})if__name____main__:main()推理测试代码predict_cdw.pyfromultralyticsimportYOLOimportcv2# 加载训练最优权重modelYOLO(runs/train/cdw_material_detect/weights/best.pt)if__name____main__:img_pathcdw_test.jpgresultsmodel.predict(sourceimg_path,conf0.25,iou0.45,saveTrue)result_imgresults[0].plot()cv2.imshow(CDW Materials Detection,result_img)cv2.waitKey(0)cv2.destroyAllWindows()核心标签#建筑垃圾检测#固废识别#建材分类#YOLO数据集#工地物料识别#航拍垃圾巡检
