In order to solve the problem that YOLOv5 algorithm is not accurate in locating the target object and local features

a YOLOv5-ABN target detection network based on YOLOv5 is proposed.Firstly

in order to get accurate positioning information

this paper combines spatial attention mechanism and coordinate attention module in convolution attention mechanism.A spatial and coordinated attention module is proposed

which integrates spatial feature information and coordinate information.The spatial and coordinated attention module is added to the Backbone network of YOLOv5.The weighted bidirectional feature pyramid network is introduced into the Neck detection layer

and the staggered box parameters are optimized

which improves the detection accuracy and positioning ability of the network for sample models and local features.The customized data set and Cornell data set are trained with the improved algorithms respectively.The experiment results show that the improved algorithm is better than YOLOv5 algorithm

and the target object is more accurate and more robust without reducing the recognition accuracy.