针对分选线上樱桃叠果和表面瑕疵难以准确实时分割的问题,该研究提出了一种基于改进YOLOv8n-seg的多分支实时分割模型,命名SemIns-YOLOv8。该模型在YOLOv8n-seg的PAN-FPN(path aggregation network for feature pyramid network)模块后...针对分选线上樱桃叠果和表面瑕疵难以准确实时分割的问题,该研究提出了一种基于改进YOLOv8n-seg的多分支实时分割模型,命名SemIns-YOLOv8。该模型在YOLOv8n-seg的PAN-FPN(path aggregation network for feature pyramid network)模块后引入基于上下文集成的语义分割模块,并采用交叉熵损失与Dice Loss联合的损失函数,替代原始实例分割模块对果梗和瑕疵进行识别,既提升了尺寸较小及特征不明显瑕疵的识别精度,又缩短了图像的识别时间。同时,通过提高特征图分辨率并引入豪斯多夫距离损失(Hausdorff distance loss,HD Loss)构建边界特征增强的实例分割模块,实现了樱桃重叠果体的精准分离。试验结果表明,SemIns-YOLOv8在樱桃分割任务中果体mAP50-95(mean average precision at intersection over union thresholds from 0.50 to 0.95)、果梗IoU和瑕疵mIoU(mean intersection over union)分别为98.20%、92.15%和65.97%,与YOLOv8n-seg相比,提升了2.10、2.33和14.35个百分点,并且在模型输入尺寸为1024×384像素时,单帧推理时间为23 ms,可为线上水果外观品质实时分选提供参考。展开更多
The hydrological character parameters of the litter amount, the maximum water capacity and water absorption speed of the litter on the forest land under four type of \%Abies fabri\% forest (young, half_mature,mature a...The hydrological character parameters of the litter amount, the maximum water capacity and water absorption speed of the litter on the forest land under four type of \%Abies fabri\% forest (young, half_mature,mature and overmature stand)on the Gongga Mountain were studied in this paper. The result showed that the litter amount under the overmature stand was the maximum(67 8?t·hm -2 ), the second was the mature stand(53 4 t·hm -2 ), the third was the half mature stand(40 4 t·hm -2 ), the fourth was the young stand(32 6?t·hm -2 ). The maximum water capacity of the undecomposed litter under the mature stand was 428 8% of the litter’s dry weight, that of the litter under the overmature stand was 411 6%, that of the litter under the half_mature stand was 378 8%, that of the litter under the young stand was 296 7%. The maximum water capacity of the half_decomposed litter under the mature stand was 234 5% of the litter’s dry weight, that of the litter under the overmature stand was 175 6%, that of the litter under the half mature stand was 160 3%, that of the litter under the young stand was 132 5%. The equation between the water absorption speed of the litter and the soaked time is S=kt n. The result also shows that the litter’s variation with the time has no relation to forest age, while its decomposition degree is positively related to stand age.展开更多
文摘针对分选线上樱桃叠果和表面瑕疵难以准确实时分割的问题,该研究提出了一种基于改进YOLOv8n-seg的多分支实时分割模型,命名SemIns-YOLOv8。该模型在YOLOv8n-seg的PAN-FPN(path aggregation network for feature pyramid network)模块后引入基于上下文集成的语义分割模块,并采用交叉熵损失与Dice Loss联合的损失函数,替代原始实例分割模块对果梗和瑕疵进行识别,既提升了尺寸较小及特征不明显瑕疵的识别精度,又缩短了图像的识别时间。同时,通过提高特征图分辨率并引入豪斯多夫距离损失(Hausdorff distance loss,HD Loss)构建边界特征增强的实例分割模块,实现了樱桃重叠果体的精准分离。试验结果表明,SemIns-YOLOv8在樱桃分割任务中果体mAP50-95(mean average precision at intersection over union thresholds from 0.50 to 0.95)、果梗IoU和瑕疵mIoU(mean intersection over union)分别为98.20%、92.15%和65.97%,与YOLOv8n-seg相比,提升了2.10、2.33和14.35个百分点,并且在模型输入尺寸为1024×384像素时,单帧推理时间为23 ms,可为线上水果外观品质实时分选提供参考。
文摘The hydrological character parameters of the litter amount, the maximum water capacity and water absorption speed of the litter on the forest land under four type of \%Abies fabri\% forest (young, half_mature,mature and overmature stand)on the Gongga Mountain were studied in this paper. The result showed that the litter amount under the overmature stand was the maximum(67 8?t·hm -2 ), the second was the mature stand(53 4 t·hm -2 ), the third was the half mature stand(40 4 t·hm -2 ), the fourth was the young stand(32 6?t·hm -2 ). The maximum water capacity of the undecomposed litter under the mature stand was 428 8% of the litter’s dry weight, that of the litter under the overmature stand was 411 6%, that of the litter under the half_mature stand was 378 8%, that of the litter under the young stand was 296 7%. The maximum water capacity of the half_decomposed litter under the mature stand was 234 5% of the litter’s dry weight, that of the litter under the overmature stand was 175 6%, that of the litter under the half mature stand was 160 3%, that of the litter under the young stand was 132 5%. The equation between the water absorption speed of the litter and the soaked time is S=kt n. The result also shows that the litter’s variation with the time has no relation to forest age, while its decomposition degree is positively related to stand age.
