CTSegFormer:一种面向烟丝组分占比测定的语义分割网络模型
CTSegFormer: a semantic segmentation network for cut tobacco component proportion measurement
-
摘要: 【目的】提升对烟丝组分占比测定的效率和准确性。【方法】提出一种语义分割模型CTSegFormer,以SegFormer为基础骨架,设计多类型卷积并行融合、增强的高效加性注意力机制及渐进式深度监督模块,完成对叶丝、薄片丝、梗丝、膨胀丝4类烟丝及梗签的有效识别和分割,在此基础上实现对烟丝组分占比的准确测定,并构建烟丝组分测定数据集用于测试与验证。【结果】与现有语义分割模型相比,CTSegFormer模型在分割和识别烟丝能力上的mIoU指标最高,同时,在测定多个不同组分配比样品时,其占比预测相对误差不超过6%。【结论】提出的CTSegFormer模型能较好地适用于烟丝组分占比测定任务,对监控成品烟支品质有重要意义。Abstract: 【Objective】To improve the efficiency and accuracy of component proportion measurement for tobacco shreds.【Methods】A semantic segmentation model named CTSegFormer is proposed. Taking SegFormer as the backbone, the model designs parallel fusion of multiple types of convolutions, an enhanced efficient additive attention mechanism, and a progressive deep supervision module. It achieves effective recognition and segmentation of four categories of tobacco shreds including lamina shreds, reconstituted tobacco shreds, stem shreds, expanded shreds, as well as stem sticks. On this basis, accurate measurement of tobacco shred component proportions is realized, and a tobacco shred component measurement dataset is constructed for testing and validation.【Results】Compared with existing semantic segmentation models, CTSegFormer achieves the highest mIoU in tobacco shred segmentation and recognition. Meanwhile, the relative error of proportion prediction is no more than six percent when testing multiple samples with different component ratios.【Conclusion】The proposed CTSegFormer model can be well applied to the task of tobacco shred component proportion measurement, and it is of great significance for monitoring the quality of finished cigarettes.
-
Key words:
- tobacco shred /
- component proportion /
- semantic segmentation /
- SegFormer
-
-
[1]
Gao Z Y,Wang A,Ding H,et al.Tobacco material composition recognition method based on convolutional neural network[J].Tobacco Science & Technology,2017,50(9):68-75.
-
[2]
王小明,魏甲欣,马飞,等.基于AdaBoost集成学习的烟丝组分识别[J].食品与机械,2022,38(3):205-211.
Wang X M,Wei J X,Ma F,et al.Identification of cut tobacco components based on AdaBoost ensemble learning[J].Food & Machinery,2022,38(3):205-211. -
[3]
刘栋,陈越立,李华杰,等.取样量、取样次数对烟丝混合均匀性检测的影响[J].郑州轻工业学院学报(自然科学版),2013(1):45-49. Liu D,Chen Y L,Li H J,et al.Influence of sampling volume and frequency on the mixing uniformity test of cut filler[J].Journal of Zhengzhou University of Light Industry (Natural Science),2013
(1):45-49. -
[4]
-
[5]
寇霄腾,张勇,张卉,等.基于RGB图像特征的卷烟梗丝掺配比例检测[J].食品与机械,2021,37(9):78-82.
Kou X T,Zhang Y,Zhang H,et al.Detection of blending proportion of cut tobacco stem based on RGB image features[J].Food & Machinery,2021,37(9):78-82. -
[6]
杨天卓,何晋,吴恋恋,等.基于高光谱成像的烟丝掺配比例检测研究[J].轻工学报,2025,40(3):115-126.
Yang T Z,He J,Wu L L,et al.Detection of tobacco blend ratio based on hyperspectral imaging[J].Journal of Light Industry,2025,40(3):115-126. -
[7]
刘松岳,俞世康,赵宇,等.基于深度语义分割的重叠烟叶分级方法[J].计算机与数字工程,2023,51(7):1645-1650.
Liu S Y,Yu S K,Zhao Y,et al.Overlapping tobacco ranking method based on deep semantic segmentation[J].Computer and Digital Engineering,2023,51(7):1645-1650. -
[8]
Dosovitskiy A,Beyer L,Kolesnikov A,et al.An image is worth 16x16 words:transformers for image recognition at scale[EB/OL].V2.arXiv(2021-06-03)[2025-06-30].https://doi.org/10.48550/arxiv.2010.11929.
-
[9]
Vaswani A,Shazeer N,Parmar N,et al.Attention is all you need[C]//Advances in Neural Information Processing Systems,2017,30:5998-6008.
-
[10]
Xie E Z,Wang W H,Yu Z D,et al.SegFormer:Simple and efficient design for semantic segmentation with transformers[EB/OL].V3.arXiv(2021-10-28)[2025-06-30].https://doi.org/10.48550/arxiv.2105.15203.
-
[11]
高延海.基于改进SegFormer的语义分割方法的研究[D].青岛:青岛理工大学,2024. Gao Y H.Research on semantic segmentation method based on improved SegFormer[D].Qingdao:Qingdao University of Technology,2024.
-
[12]
王晓杰,陈少康,闫皓炜,等.融合注意力机制的SegFormer遥感影像道路识别[J].计算机系统应用,2024,33(11):186-193.
Wang X J,Chen S K,Yan H W,et al.Road recognition in remote sensing images using SegFormer fused with attention mechanism[J].Computer Systems and Applications,2024,33(11):186-193. -
[13]
胡涛涛,李屹旭,张俊.基于注意力机制改进的SegFormer遥感图像语义分割方法[J].激光杂志,2024,45(7):130-136.
Hu T T,Li Y X,Zhang J.Improved SegFormer network semantic segmentation method for remote sensing images based on attention mechan[J].Laser Journal,2024,45(7):130-136. -
[14]
田雪伟,汪佳丽,陈明,等.改进SegFormer网络的遥感图像语义分割方法[J].计算机工程与应用,2023,59(8):217-226.
Tian X W,Wang J L,Chen M,et al.Improved SegFormer network based method for semantic segmentation of remote sensing images[J].Computer Engineering and Applications,2023,59(8):217-226. -
[15]
蒋仕新,唐椿程,杨建喜,等.基于改进SegFormer的混凝土桥梁表观病害轻量级识别方法[J].中国公路学报,2024,37(2):77-87.
Jiang S X,Tang C C,Yang J X,et al.Lightweight detection method of surface damage on concrete bridges based on improved SegFormer[J].China Journal of Highway and Transport,2024,37(2):77-87. -
[16]
张铭泉,张泽恩,曹锦纲,等.结合SegFormer与增强特征金字塔的文本检测方法[J].智能系统学报,2024,19(5):1111-1125.
Zhang M Q,Zhang Z E,Cao J G,et al.Text detection method combining SegFormer with an enhanced feature pyramid[J].CAAI Transactions on Intelligent Systems,2024,19(5):1111-1125. -
[17]
Yu Z T,Zhao C X,Wang Z Z,et al.Searching central difference convolutional networks for face anti-spoofing[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Seattle,WA,USA.IEEE,2020:5294-5304.
-
[18]
Chen L C,Zhu Y K,Papandreou G,et al.Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Computer Vision-ECCV 2018.Cham:Springer,2018:833-851.
-
[19]
Shaker A,Maaz M,Rasheed H,et al.SwiftFormer:efficient additive attention for transformer-based real-time mobile vision applications[C]//2023 IEEE/CVF International Conference on Computer Vision (ICCV).Paris,France.IEEE,2023:17379-17390.
-
[20]
Ronneberger O,Fischer P,Brox T.U-Net:convolutional networks for biomedical image segmentation[C]//Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015.Cham:Springer,2015:234-241.
-
[21]
Badrinarayanan V,Kendall A,Cipolla R.SegNet:a deep convolutional encoder-decoder architecture for image segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(12):2481-2495.
-
[22]
Li B Y, Liu Y, Wang X G,et al.Gradient harmonized single-stage detector[C].//Proceedings of the AAAI Conference on Artificial Intelligence,2019,33(1):8577-8584.
-
[23]
Wang W H,Xie E Z,Li X,et al.PVT v2:improved baselines with pyramid vision transformer[J].Computational Visual Media,2022,8(3):415-424.
-
[24]
Mehta S,Rastegari M.Separable self-attention for mobile vision transformers[PP/OL].arXiv(2022-06-06)[2025-06-30].https://doi.org/10.48550/arxiv.2206.02680.
-
[25]
Zheng S X,Lu J C,Zhao H S,et al.Rethinking semantic segmentation from a sequence-to-sequence perspective with transformers[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Nashville,TN,USA.IEEE,2021:6877-6886.
-
[26]
Long J,Shelhamer E,Darrell T.Fully convolutional networks for semantic segmentation[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston,MA,USA.IEEE,2015:3431-3440.
-
[27]
Zhao H S,Shi J P,Qi X J,et al.Pyramid scene parsing network[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu,HI,USA IEEE,2017:6230-6239.
-
[28]
Sun k,Xiao b,Liu D,et al.Deep high-resolution representation learning for human pose estimation[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Long Beach,CA,USA.IEEE,2019:5686-5696.
-
[29]
Cheng m m,Guo m h,Hou Q B,et al.SegNeXt:rethinking convolutional attention design for semantic segmentation[C]//Advances in Neural Information Processing Systems 35.New Orleans,Louisiana,USA.Neural Information Processing Systems Foundation,Inc.(NeurIPS),2022:1140-1156.
-
[30]
周博,朱文魁,王赵改,等.基于太赫兹时域光谱技术的烟草组分识别[J].农业工程学报,2022,38(10):310-316.
Zhou B,Zhu W K,Wang Z G,et al.Identification of tobacco materials based on terahertz time-domain spectroscopy[J].Transactions of the Chinese Society of Agricultural Engineering,2022,38(10):310-316.
-
[1]
-
点击查看大图
计量
- PDF下载量: 5
- 文章访问数: 283
- 引证文献数: 0
下载:
