深度可分离卷积神经网络在自动分拣中的应用

针对传统的花卉分类算法在工业自动化分拣应用中出现模型参数过大、分拣精度不高的问题，提出一种基于深度学习的花卉识别算法。介绍了花卉分类算法在工业花卉包装分拣系统中的应用；根据实际需求，采用一种深度可分离卷积神经网络提取花卉特征，并详细分析了网络的模型结构；为了提高模型训练速度，提出了一种微调的模型训练方法。实验结果表明，所采用的花卉分类算法在工业花卉自动分拣的应用中相比传统算法，准确率更高、稳定性更好、应用更加广泛。

Depth Separable Convolutional Neural Network and Its Application in Automatic Sorting

Aimed at the problem of too large model parameters and low sorting precision of the traditional flower classification algorithm in industrial automation sorting application, a flower recognition algorithm based on deep learning was proposed. The application of flower classification algorithm in industrial flower packaging sorting system was introduced. According to the actual demand, a deep separable convolutional neural network was used as the flower feature extraction, and the model structure of the network was analyzed in detail. In order to improve the speed of model training, a fine-tuned model training method was proposed. The experimental results showed that the flower classification algorithm used in the industrial flower automatic sorting application had higher accuracy, better stability and wider application than traditional algorithms.