Feature enhanced target detection method based on feature pyramid network

Abstract

The invention discloses an image target detection method based on a feature pyramid network. A top-down module TDM, a bottom-up module BUM and a fusion expansion module FEM are added to a backbone network model adopted before the backbone network model is sent to a target detector; therefore, a novel enhanced feature pyramid network eFPN is formed. The detection performance of small-size objects,large-size objects and medium-size objects can be enhanced at the same time, only few parameters and calculation amount are introduced, the problem of multi-scale targets in the target detection process is better solved, the performance of a single-stage network and a double-stage network is obviously enhanced, and therefore the detection performance of the multi-scale target objects in the imageis improved.

Description

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine

Login to view more

AI Technical Summary

Benefits of technology

This patented technology describes combining different types of components together to make sure they work properly when dealing with smaller or larger items like pills or bags. It also includes advanced techniques to identify specific areas within those materials more accurately than previous methods. Overall, this innovation allows us to create powerful networks capable of identifying multiple sizes of things efficiently without compromising their effectiveness over other systems.

Problems solved by technology

This patents describes how traditional methods used for detecting targets like cars or people use convolution networks that only take advantage from certain layers of an artificial intelligence system' s trained model. These models may lose their effectiveness when dealing with complex images due to factors like noise interference caused by different levels of detail captured during training process.

Images