T Wang, L Wan, L Tang, M Liu - Applied Intelligence, 2022 - drive.google.com
To solve the problem of numerous deep convolutions in YOLOv4, which generates many
redundant background features so that it cannot focus on pedestrians at a specific scale, we …