Deep learning-based image recognition model training method and system

The invention relates to the technical field of computers, and relates to an image recognition model training method and system based on deep learning, and the method comprises the steps: selecting a pre-training sample to carry out the pre-training of an image recognition model; calculating a multi-classification evaluation index of a plurality of classification targets corresponding to the pre-trained image recognition model, and setting an acquisition probability of the corresponding classification targets according to the multi-classification evaluation index; detecting the samples in the training sample set by using the pre-training model to determine the loss value and the prediction classification of the samples, determining the focus attention of the samples according to the loss value of the samples and the consistency of the prediction classification and the real classification, calculating the selection probability of the samples, and selecting the samples according to the selection probability to