摘 要：针对目前槟榔点卤工艺中卤水量不好精确控制的问题，文章提出采用深度学习的方式对槟榔内轮廓进行语义分割，分离出内轮廓并计算出相应面积，最后推算出比较准确的卤水量。其中，网络模型以 UNet 为基础模型，考虑到模型的通用性，将 UNet 的 encoder 特征提取部分替换成 VGG16 网络。实验结果表明，该网络模型对于槟榔内外腔的分割效果很好，分割精度达到 97% 以上，性能优于不进行迁移学习的 UNet。
关键词：语义分割；UNet；VGG16；槟榔轮廓分割
DOI:10.19850/j.cnki.2096-4706.2023.05.036
中图分类号：TP391.4 文献标识码：A 文章编号：2096-4706（2023）05-0149-04
Application of Areca Nut Contour Image Segmentation Algorithm Based on Deep Learning
CHENG Pan
(Sankyo-HZ Precision Co., Ltd., Huizhou 516006, China)
Abstract: Aiming at the problem that the brine amount is not well controlled accurately in the process of adding brine to areca nut at present, this paper proposes to use the deep learning method to perform semantic segmentation on the inner contour of areca nut, after separating the inner contour and calculating the corresponding area, and it finally calculates the more accurate brine amount. The network model is based on UNet model. Considering the universality of the model, the encoder feature extraction part of UNet is replaced by VGG16 network. The experimental results show that the network model has a good segmentation effect for the internal and external cavities of areca nut, with the segmentation accuracy of more than 97%, and its performance is better than that of UNet without migration learning.
Keywords: semantic segmentation; UNet; VGG16; areca nut contour segmentation
参考文献：
[1] LECUN Y，BOTTOU L，BENGIO Y，et al. Gradient-based learning applied to document recognition [J].Proceedings of the IEEE， 1998，86（11）：2278-2324.
[2] KRIZHEVSKY A，SUTSKEVER I，HINTON G. ImageNet Classification with Deep Convolutional Neural Networks [J].Advances in neural information processing systems，2012，25（2）：75-79.
[3] TAIGMAN Y，YANG M，RANZATO M，et al. DeepFace： Closing the Gap to Human-Level Performance in Face Verification [C]//2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus：IEEE，2014：1701-1708.
[4] SIMONYAN K，ZISSERMAN A. Very Deep Convolutional Networks for Large-Scale Image Recognition [J/OL].arXiv：1409.1556[cs.CV].（2015-04-10）.https://arxiv.org/abs/1409.1556.
[5] 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.
[6] SHELHAMER E，LONG J，D A R R E L L T. F u l l y Convolutional Networks for Semantic Segmentation [J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2016，39（4）： 640-651.
[7] IGLOVIKOV V，SHVETS A. TernausNet：U-Net with VGG11 Encoder Pre-Trained on ImageNet for Image Segmentation [J/OL].arXiv：1801.05746 [cs.CV].（2018-01-17）.https://arxiv.org/abs/1801.05746.
[8] MILLETARI F，NAVAB N，AHMADI S A. V-Net： Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation [C]//2016 Fourth International Conference on 3D Vision （3DV）.Stanford：IEEE，2018：565-571.
[9] ZHOU Z，SIDDIQUEE M，TAJBAKHSH N，et al. UNet++：A Nested U-Net Architecture for Medical Image Segmentation [C]//DLMIA 2018，ML-CDS 2018.Cham：Springer，2018：3-11.
[10] NAWAZ A，AKRAM U，SALAM A，et al. VGG-UNET for Brain Tumor Segmentation and Ensemble Model for Survival Prediction [C]//2021 International Conference on Robotics and Automation in Industry （ICRAI）.Rawalpindi：IEEE，2021：1-6.
作者简介：程盼（1988—），男，汉族，湖北天门人，高级工程师，硕士，研究方向：机器视觉。