Addressing the issue of poor recognition of manually painted serial numbers in steel ladle text recognition.We propose a classifica-tion and recognition method based on deep learning to identify the serial numbers.By utilizing data augmentation techniques such as image transformations,adding Gaussian noise,Cutout,Gridmask,and other image occlusion methods,we have improved the accuracy and generalization capabilities of the trained model.The experiments show that the method achieves high accuracy in identifying steel ladle serial numbers in real scene scenarios.
Deep learningClassification recognitionData augmentationLadle number
万方数据
