Saliency perception image cutting method and device, computing equipment and storage medium

The invention relates to a saliency perception image cutting method and device, computing equipment and a storage medium, the saliency perception image cutting method comprises cutting model training and image cutting, and the cutting model training comprises the steps of creating an initial cutting image based on a saliency region of an original image, and performing image cutting based on the initial cutting image, generating a plurality of candidate cutting pictures in a grid anchor mode, then extracting features of an original image, and cutting the extracted image features by using a saliency-oriented region-of-interest pooling operator and a saliency-oriented discarded region pooling operator to obtain an alignment feature graph of the region-of-interest and the discarded region; and finally, feeding the alignment feature map into an aesthetic evaluation network capable of predicting the aesthetic score of the cut image for training, and obtaining a completely trained cut model. According to the method