Logo detection method and device based on YOLOv7 model

• Main Authors: LI YEHENG, SU QIANMIN, FANG ZHIJUN, LUO GUANGSHENG
• Format: Patent
• Language: Chinese; English
• Published: 05.04.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses a Logo detection method and device based on a YOLOv7 model, and the method comprises the steps: constructing a Logo detection model employing the YOLOv7 model as a basic framework, and a backbone network comprises a first feature extraction module and three groups of second feature extraction modules which are sequentially connected; the second feature extraction module comprises a CSMP module and an ELAN module; feature extraction based on a channel attention mechanism and feature extraction based on a space attention mechanism are conducted on the first feature maps after maximum pooling and average pooling through a CSMP module of a first set of second feature extraction modules, and a channel attention feature map and a space attention feature map are obtained; splicing the channel attention feature map and the space attention feature map to obtain an enhanced feature map, and processing the enhanced feature map through an ELAN module to obtain a second feature map; and processing