Pedestrian　detection　aims　to　automatically　identify　and　locate　pedestrian　objects　in　images　or　videos,　enabling　applications　such　as　intelligent　surveillance,　traffic　safety,　and　crowd　counting.　To　address　the　issue　of　low　accuracy　in　pedestrian　detection　using　the　YOLOv4　object　detection　algorithm,　an　improved　YOLOv4　algorithm　is　proposed　to　enhance　pedestrian　detection　performance.　By　incorporating　the　DenseNet　model　and　ECANet　attention　mechanism,　experiments　are　conducted　on　the　INRIA　dataset.　The　improved　YOLOv4　algorithm　achieves　an　average　precision　(AP)　of　93.97%,　which　is　a　4.85%　improvement　compared　to　the　original　YOLOv4　algorithm.　©　2023　IEEE.