联合损失监督的高频工件深度学习识别算法

High Frequency Workpiece Deep Learning Recognition Algorithm Based on Joint Loss Supervision

针对高频工件种类多、类间相似度较高造成的识别准确率低的问题,提出一种联合损失监督的深度学习识别算法。搭建基于卷积神经网络的图像特征向量编码模型,采用角度余量损失替换SoftMax损失,以减小工件类内特征之间的距离,完成同类工件的鲁棒性表示;引入隔离损失以增大异类工件特征之间的距离,实现异类工件的良好性区分。实验结果表明:该方法相较于传统的图像识别方法,识别准确率更高;相较于单一的角度余量和隔离损失,识别准确率分别提高了3.97%和13.88%。

To improve the low recognition accuracy caused by wide varieties of high frequency artifacts and high similarity between classes,a deep learning algorithm with joint loss supervision is proposed.An image feature vector encoding model is built based on convolutional neural network,and the SoftMax loss is replaced by the angle margin loss to reduce the distance between the features within the workpiece class and complete the robust representation of similar workpieces.The isolation loss is introduced to increase the distance between the features of heterogeneous workpieces and achieve good discrimination of heterogeneous workpieces.The experimental results show that the recognition accuracy of the proposed method is higher than that of the traditional image recognition method,with the single angle margin increasing by 3.97%and isolation loss 13.88%respectively.