Multi-class target object detection method and device, electronic device and storage medium

Abstract

The invention provides a multi-class target object detection method and device, an electronic device, a storage medium. The multi-class target object detection method comprises: training a first neural network model to enable the first neural network model to output a quasi-first target frame and a quasi-second target frame, the quasi-first target frame indicating a first class of target objects,and the quasi-second target frame indicating a second class of target objects; training a second neural network model to enable the second neural network model to output second-level confidence coefficients and attribute information of the quasi-first target frame and the quasi-second target frame; taking the trained first neural network model and the trained second neural network model as a multi-class target object detection model; and inputting the to-be-detected picture into the multiple types of target object detection models, and determining the superiority according to the second-levelconfidence coefficients and attribute information of the quasi-first target box and the quasi-second target box output by the multiple types of target object detection models. According to the methodand the equipment provided by the invention, simultaneous detection of multiple types of objects is realized.

Description

technical field [0001] The invention relates to the field of computer applications, in particular to a multi-type target object detection method, device, electronic equipment, and storage medium. Background technique [0002] With the popularization of safe cities and the rapid growth of social surveillance, based on the development of deep learning, AI (Artificial Intelligence), and various technical requirements for analyzing various surveillance pictures and surveillance videos, it is possible to accurately and quickly from Detecting specific targets in pictures and videos, especially human faces and pedestrians, has become an extremely important key part. It is necessary to ensure high detection accuracy and not miss any targets, which has improved a solid foundation for subsequent analysis. [0003] Pedestrian and face detection based on traditional algorithms are independent modules, and require separate design of features and separate training of classifiers, which ca...

AI Technical Summary

Problems solved by technology

[0003] Pedestrian and face detection based on traditional algorithms are independent modules, and require separate design of features and separate training of classifiers, which can guarantee the detection rate of faces and pedestrians in some specific scenes. Accurate and fast detection in pictures and videos of various complex and diverse scenes is relatively difficult, and qualified detection results cannot be obtained

Images