Image recognition method and device

The invention relates to an image recognition method and device. The method comprises the following steps: establishing a luggage case image recognition model based on a convolutional neural network or a Transform neural network; based on supervised learning and unsupervised clustering contrast learning, a small amount of labeled training data and a large amount of unlabeled training data are utilized to train the model. The luggage case image is used as the unique identity authentication of luggage case identification, the characteristics of the luggage case image are analyzed and bound to the corresponding passenger, the luggage case image of the passenger is identified, and the retrieval and tracking of the luggage case image of the passenger in a civil aviation scene are realized. 本发明涉及一种图像识别的方法和装置。该方法基于卷积神经网络或Transformer神经网络，建立行李箱图像识别模型；基于有监督学习和无监督聚类对比学习，利用少量有标注和大量无标注训练数据训练模型。以此将行李箱图像作为行李箱识别的唯一身份认证，分析行李箱图像特征并绑定到对应的旅客，对旅客行李箱图像进行识别，实现在民航场景下，旅客行李箱图像的检索和跟踪。