Image recognition method and device, equipment, storage medium and program product

The embodiment of the invention provides an image recognition method and device, equipment, a storage medium and a program product. A method of training a model for image recognition is provided. The method includes determining, using a model, a predicted category of the image based on a first feature representation of the image, the predicted category selected from a plurality of candidate known categories. The method further includes determining, using the model, a distance between a second feature representation of the image in a feature space for open set identification and a center vector for the tag category or the pseudo tag category based on the tag category or the pseudo tag category of the image. The method further includes training a model based on the predicted category, the tag category or the pseudo tag category of the image, and the distance. In this way, the model can be efficiently trained to utilize the model to perform open set recognition on the image while determining the prediction category of the image. In some embodiments, in the field of chip design tools, the image recognition scheme disclosed by the invention can efficiently execute wafer defect recognition so as to improve the production efficiency of chips.