Target identification method and system based on YOLOv7 model, electronic equipment and medium

• Main Authors: GU SIRUI, ZHAO FANG, REN XIAOTIAN, ZUO GUANFANG, WANG SICHENG, CHENG KUN, RUAN YIYANG
• Format: Patent
• Language: Chinese; English
• Published: 31.10.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses a target recognition method and system based on a YOLOv7 model, electronic equipment and a medium. The method comprises the steps that S1, a to-be-recognized junk image set is acquired; s2, a YOLOv7 model is improved, wherein the YOLOv7 model comprises a Backbone network, a Neck network and a Head network which are connected in sequence; in a Backbone network, a multi-branch stacking module ELAN-GR is used for replacing a stacking module ELAN; the method comprises the following steps: adding two sequentially connected strong global attention combination modules CMH behind a first multi-branch stacking module ELAN-GR of a Backbone network, wherein the two sequentially connected strong global attention combination modules CMH serve as a feature map of strong position information to generate a branch B1; s3, training the improved YOLOv7 model by adopting a junk image set to be identified; and S4, inputting a to-be-detected image into the improved YOLOv7 model obtained by training to obtai