A method, device and computer recording medium for target labeling of images

Abstract

The present disclosure provides a method, device, and computer recording medium for tagging images with targets. The method includes: obtaining a trained first neural network for classifying images related to targets. The first neural network includes a multi-layer volume Convolutional neural network and fully connected layer, each layer of convolutional neural network includes convolutional layer, activation function layer and downsampling layer in sequence; the image is processed by the processor using the obtained multi-layer convolutional neural network of the first neural network processing; the processor performs large-value space position sampling on the output of at least one layer of the activation function layer of the first neural network; the processor maps the output of the down-sampling layer of the corresponding layer to an image based on the result of the large-value space position sampling A mapping matrix is ​​obtained in the coordinate space; the processor obtains a target position mask based on the mapping matrix; and the processor marks the target based on the target position mask. The training samples of this method are easier to obtain and the training is easier.

Description

technical field [0001] The present disclosure relates to the field of image processing, and in particular to a method, device and computer recording medium for object labeling on images. Background technique [0002] The location of lesions in medical images needs to be determined by experienced doctors, and the accuracy of determining the location of lesions based on traditional feature detection or pattern recognition algorithms is low. At present, although machine learning or deep learning can be used to obtain algorithms for automatically labeling lesion locations with high accuracy, these algorithms need to be trained with images that have labeled lesion locations, and lesion location information images are difficult to obtain in reality. Contents of the invention [0003] Aiming at the above-mentioned technical problems existing in the prior art, the present disclosure provides a method, device and computer recording medium for tagging images with objects, through a...

AI Technical Summary

Problems solved by technology

At present, although machine learning or deep learning can be used to obtain algorithms for automatically labeling lesion locations with high accuracy, these algorithms need to be trained with images that have labeled lesion locations, and lesion location information images are difficult to obtain in reality

Images