基于Opt-MobileNetV3的大豆种子异常籽粒识别研究

Abnormal Soybean Grains Recognition Based on Opt-MobileNetV3

针对大豆异常籽粒识别模型参数量过大、计算成本高、准确率较低等问题,提出了一种改进的轻量级神经网络MobileNetV3模型,将其层数减少,加快模型的训练和推理速度,增加全连接层和Softmax层以增加模型的非线性判别能力以及利于多分类任务的输出,使用全局平均池化代替全局最大池化减少信息丢失,通过添加Dropout层以及去掉MobileNetV3中SE Block注意力机制来增加模型的泛化能力。试验结果表明:将大豆籽粒图像数据经过传统的卷积神经网络AlexNet、VGG16与轻量级神经网络MobilenetV3训练测试结果进行对比,AlexNet算法最终平均精度均值(Mean average precision,mAP)为87.3%、VGG16算法为87.7%,二者mAP相差较小,但两者在训练过程中模型内存占用量及训练时间相差较大,其中AlexNet模型内存占用量为7070 kB,训练时间为5420.59 s,而VGG16模型内存占用量为19674 kB,训练时间为8282.68 s,整体来看AlexNet相对更好。通过对轻量级神经网络MobileNetV3模型的识别训练,最终模型内存占用量为32153 kB,训练时间为6298.29 s,mAP达到90.6%,相比两个传统算法更高,更适合大豆异常籽粒的分类识别。为了提高训练精度及速度,通过对MobileNetV3网络模型结构调整改进,最终优化改进后的Opt-MobileNetV3网络模型mAP达到95.7%,相较传统MobileNetV3神经网络mAP提高5.1个百分点,模型内存占用量为9317 kB,减小22836 kB,同时训练时间节省696.57 s。优化后的模型实现了模型减小、准确率提高、训练速度加快,可完成大豆异常籽粒识别任务。

In response to the problems of excessive parameter quantity,high computational cost,and low accuracy in the recognition model of soybean abnormal seeds,an improved lightweight neural network MobileNetV3 model was proposed,which reduced the number of layers,accelerated the training and inference speed of the model,increased the nonlinear discrimination ability of the model by adding fully connected layers and softmax layers,and facilitated the output of multiple classification tasks,by using global average pooling instead of global maximum pooling to reduce information loss,and increasing the model’s generalization ability by adding a Dropout layer and removing the SE Block attention mechanism in MobileNetV3.The experimental results showed that after comparing the soybean seed image data with the traditional convolutional neural networks AlexNet,VGG16,and lightweight neural network MobilenetV3,the AlexNet algorithm’s final mean average precision(mAP)was 87.3%,and the VGG16 algorithm’s mAP was 87.7%.The difference in mAP between the two was small,but there was a significant difference in model size and training time during the training process,the AlexNet model had a model size of 7070 kB and a training time of 5420.59 s,while the VGG16 model had a model size of 19674 kB and a training time of 8282.68 s.Overall,AlexNet was relatively better.The recognition and training of the lightweight neural network MobileNetV3 model resulted in a model size of 32153 kB,a training time of 6298.29 s,and an mAP of 90.6%,which was higher than that of the two traditional algorithms and more suitable for the classification and recognition of abnormal soybean seeds.In order to improve training accuracy and speed,the structure of the MobileNetV3 network model was adjusted and improved.The optimized Opt-MobileNetV3 network model mAP reached 95.7%,which was 5.1 percentage points higher than that of the traditional MobileNetV3 neural network mAP.The model size was 9317 kB,reduced by 22836 kB,and training time was saved by 696.57 s.The optimized model achieved reducing model size,improving accuracy,and faster training speed,which can meet the task of identifying abnormal soybean seeds.