Micro lymph node image small target detection method based on attention mechanism

The invention discloses a tiny lymph node image small target detection method based on an attention mechanism. The method comprises the following steps: carrying out preprocessing adjustment on an original tiny lymph node area image, carrying out matrix conversion on all the images, generating a feature information matrix, then carrying out iterative conversion, summing loss values, averaging, and optimizing parameters in a training model; comparing the result of the training model with the manual calibration result to obtain the average precision of all the images in the verification set, and extracting the highest precision as the optimal training model parameter; and sending the obtained feature information matrix into the optimal training model to obtain an automatically calibrated detection image, judging whether a lesion area exists or not according to the position of the anchor frame, and obtaining the position of the lesion area. The method has the advantages that the image can be globally analyzed, d