Problem of object classification

[zhangjy328]

Dear Author, I would like to ask: Is this model only capable of training on data with a fixed number and type of items? If each image contains a different number and type of items, and even includes duplicate objects, how should the training be conducted? From my observation of the training model, it seems that each image in the dataset must contain a fixed number of items, and each object appears exactly once. Is this the case?