Continuous learning image classification method based on double-branch network

The invention discloses a continuous learning image classification method based on a double-branch network, and belongs to the technical field of image processing. The method comprises the following steps: constructing a double-branch network which comprises a main branch and a memory branch; learning the first batch category by using the memory branch, and generating a prediction result of the first batch category; starting from the second batch, learning the t-th batch category by using the main branch, updating memory branch parameters, and generating prediction results from the first batch to the t-th batch category; and the subsequent batches are learned by adopting a method which is the same as that in the step 3 until all the batches are learned. The method can effectively avoid old category forgetting caused by new category learning and effectively overcome the prediction prejudice phenomenon caused by category imbalance, and does not additionally increase the model parameter quantity or the network s