一种改进的Mobile Net V2生活垃圾图像分类算法

An Improved Mobile Net V2 Algorithm for Refuse Classification

垃圾分类有利于资源的回收利用和减少环境污染。文章结合计算机视觉与垃圾分类任务,提出一种针对改进的Mobile Net V2生活垃圾图像分类算法。与原有的分类算法相比,改进的Mobile NetV2算法引入注意力机制引导模型关注图像的关键信息特征,提高模型特征表达的能力。此外,通过构建垃圾分类的数据集便于模型训练和测试,并提出一种类内随机图像融合的数据噪声增强策略,进一步增加数据的多样性。实验显示,改进的注意力Mobile Net V2网络进一步提升了网络的分类准确率,具备一定的实用性。

Garbage classification is beneficial to the recycling of resources and reducing environmental pollution.The article combines computer vision and garbage classification tasks,and proposes an improved Mobile Net V2 image classification algorithm for household garbage.Compared with the original classification algorithm,the improved Mobile NetV2 algorithm introduces an attention mechanism to guide the model to focus on the key information features of the image and improve the model’s ability of feature representation.In addition,it facilitates model training and testing by constructing a dataset for garbage classification,and proposes a data noise enhancement strategy of intra-class random image fusion to further increase the data diversity.The experiments show that the improved attention Mobile Net V2 network further improves the classification accuracy of the network and has certain practicality.