YOLOv11-TDSP：口腔全景片的轻量化高精度异常牙检测模型

打开文本图片集

An lightweight algorithm for multi-dimensional optimization of intelligent detection of dental abnormalities on panoramic oral x -ray images

ZHAO Taotao，NI Ming， XIA Shunxing， JIAO Yuehao，HEYating CollegeofInformationEngineering，SichuanAgriculturalUniersity，Ya'an2514，China

Abstract： Objective We propose a YOLOv1-TDSP model for improving the acuracyof dental abnormality detectionon panoramicoral X-ray images.MethodsThe SHSA single-headatention mechanism was integrated with C2PSA in the backbone layer toconstruct anew C2PSA_SHSA attention mechanism.Thecomputational redundancy wasreducedby applyingsingle-headatentiontosomeinputchannels toenhancetheeficiencyanddetectionaccuracyof the modelAsmall objectdetectionlayerwasthenintroducedintotheheadlayertocorrecttheeasilymissedandfalsedetectionsofsmallobjects. Tworoundsof structured pruning were implementedtoreducethenumberof modelparameters，avoidoverfitig，and improvetheaverageprecision.Before training，data augmentationtechniquessuchas brightness enhancementandgamma contrastdjustmentwereemployedtoenhancethegeneralizationabilityofthemodel.ResultsTheexperimentresultsshowed that the optimized YOLOv11-TDSP model achieved an accuracy of 94.5% a recall rate of 92.3% ，andanaverage precision of 95.8% fordetecting dental abnormalities.Compared with the baseline model YOLOv11n，these metrics were improved by 6.9% 7.4%， and 5.6%， respectively. The number of parameters and computational cost of the YOLOv11-TDSP model were only 12% and 13% of those of the high-precision YOLOv11x model，respectively. Conclusion The lightweight YOLOv11-TDSP model is capable of highly accurate identification of various dental diseases onpanoramic oral X-ray images. Keywords：panoramicoral x -rayimages;abnormal teethrecognition; targetdetection;YOLOv11n

第4次口腔健康流行病学调查报告，当前人们所面临的口腔健康形势极为严峻1。（剩余12840字）