Environment detection model generation method, computer equipment and readable storage medium

Abstract

The invention provides an environment detection model generation method, computer equipment and a readable storage medium, and the method comprises the steps: receiving a sample picture, and carryingout the classification of a sample in the sample picture; respectively counting the number of samples belonging to different classifications, and determining small samples in the samples according tothe number of the samples; generating a supplementary sample by using a generative adversarial network model according to the small sample, and supplementing the supplementary sample to the small sample; and forming a training set by using the samples and the supplementary samples, training the to-be-trained model by using the training set, and generating an environment detection model. Accordingto the generation method of the environment detection model provided by the invention, the generative adversarial network model is used for generating the supplementary samples to supplement the supplementary samples into the small samples so as to play a role in balancing the samples, so that the to-be-detected model can fully train each type of samples, and the accuracy and precision of the environment detection model are further improved.

Description

technical field [0001] The present invention relates to the technical field of environmental detection, in particular to a method for generating an environmental detection model, a computer device and a computer-readable storage medium. Background technique [0002] With the success of deep learning applications in the field of image and video object detection, more and more new technologies have begun to further improve the accuracy and speed of object detection. At present, target detection can be roughly divided into two categories: One-stage (single-stage) and Two-stage (two-stage). The former is represented by different versions of YOLO algorithm and SSD algorithm. Their characteristics are extracted during the detection process. The features of the candidate frame and the extracted target use the same network, or the two are combined into the same process, so it is called One-stage; the latter, the Two-stage method, is the process of extracting the candidate frame It ...

AI Technical Summary

Problems solved by technology

[0004] The present invention aims to at least solve the technical problem of low detection model accuracy caused by the imbalance of sample categories existing in the prior art

Images