结合注意力机制GoogLeNet的羊只个体识别

Recognition of sheep individual based on GoogLeNet combined with attention mechanism

随着人工智能技术的快速发展,针对羊只的智能化养殖技术需求也日益渐增。为解决羊只个体识别工作量大、工作效率低和人工依赖性强等问题,本文提出了结合注意力机制的GoogLeNet对羊只进行个体识别。模型引入通道注意力机制SENet(Squeeze-and-Excitation Networks)模块,将通道内有用信号进行筛选放大,减弱噪声的影响。在训练过程中,对学习率、优化器参数进行了调整。实验结果表明,改进网络与原始网络相比,在羊个体识别任务上具有更高的准确度以及较好的泛化性。

With the rapid development of artificial intelligence technology,the demand for intelligent sheep breeding technology is also increasing.In order to solve the problems of large workload,low work efficiency and strong artificial dependence on sheep individual identification,this paper proposes to use GoogLeNet combined with attention mechanism to identify sheep individual.This model introduces the channel attention mechanism in SE-Net(Squeeze-and-Excitation Networks)to screen and amplify the useful signals in the channel to reduce the influence of noise.During the training process,the learning rate and optimizer parameters are adjusted.The experimental results show that compared with the original network,the improved network has higher accuracy and better generalization in the sheep individual recognition task.