Generative adversarial network-based object detection method and device

Abstract

The present invention discloses a generative adversarial network-based object detection method and a device. A generative adversarial network comprises a generator and a discriminator. The method comprises the following steps that: size transformation is performed on an inputted original image, so that a first image is obtained, filtering and de-noising processing is performed on the first image, so that a second image can be obtained; the second image is input into the generative adversarial network; and the generator trains the second image to generate an object description text corresponding to the second image and transmits the object description text to the discriminator; and the discriminator discriminates whether the object description text is real data, and transmits a discrimination result to the generator, so that the generator can be adjusted and trained. The generative adversarial network-based object detection method and the device of the present invention have the advantages of high object detection efficiency and high recognition rate.

Description

technical field [0001] The invention relates to the field of target object detection, in particular to an object detection method and device based on a generative confrontation network. Background technique [0002] In the prior art, when identifying pictures, there are two existing methods. One is to use a method based on image segmentation, using image segmentation and feature extraction for identification; the other is based on artificial neural network CNN. , through the artificial neural network CNN to perform convolution calculation on the picture to detect objects. The first method is to segment the picture first and then extract the set features. This method is relatively inefficient, and the recognition rate is not high. At the same time, it needs to pre-set the types of features to be extracted, so it is very inefficient. The second method is to perform convolution on the picture to detect objects. There are repeated calculations in the CNN process, and it is nece...

AI Technical Summary

Problems solved by technology

The first method is to segment the picture first and then extract the set features. This method is relatively inefficient, and the recognition rate is not high. At the same time, it needs to pre-set the extracted feature types, so it is very inefficient.

The second method is to perform convolution on the picture to detect objects. There are repeated calculations in the CNN process, and it is necessary to set the candidate frame and extract the features of the candidate frame, which is relatively inefficient.

Images