Excuse me, can your SRC connect the deep neural network with the last fully connected layer removed to extract feature vectors (such as resnet50) for classification, and the accuracy can exceed that of the deep neural network alone? Have you ever done this kind of experiment? Can you provide the code?
Excuse me, can your SRC connect the deep neural network with the last fully connected layer removed to extract feature vectors (such as resnet50) for classification, and the accuracy can exceed that of the deep neural network alone? Have you ever done this kind of experiment? Can you provide the code?