基于深度学习的大豆豆荚类别识别研究

闫壮壮, 闫学慧, 石嘉, 孙凯, 虞江林, 张战国, 胡振邦, 蒋鸿蔚, 辛大伟, 李杨, 齐照明, 刘春燕, 武小霞, 陈庆山, 朱荣胜

Classification of soybean pods using deep learning

Zhuang-Zhuang YAN, Xue-Hui YAN, Jia SHI, Kai SUN, Jiang-Lin YU, Zhan-Guo ZHANG, Zhen-Bang HU, Hong-Wei JIANG, Da-Wei XIN, Yang LI, Zhao-Ming QI, Chun-Yan LIU, Xiao-Xia WU, Qing-Shan CHEN, Rong-Sheng ZHU

图3 深度网络模型在不同算法下的表现 a: AlexNet模型在Adam算法和SGD算法下的验证准确率; b: Vgg16模型在Adam算法和SGD算法下的验证准确率; c: GoogleNet模型在Adam算法和SGD算法下的验证精准确率; d: ResNet-50模型在Adam算法和SGD算法下的验证准确率。横轴为迭代次数, 纵轴为验证准确率。

Fig. 3 Deep network model performance under different algorithms a: the verification accuracy of the AlexNet model under the Adam algorithm and SGD algorithm; b: the verification accuracy of the Vgg16 model under the Adam algorithm and SGD algorithm; c: the verification accuracy of the GoogleNet model under the Adam algorithm and SGD algorithm; d: ResNet-50 The verification accuracy of the model under Adam algorithm and SGD algorithm. The horizontal axis is the number of iterations, and the vertical axis is the verification accuracy.