UAV reaction detection based on multi-scale feature fusion

This article first constructs a UAV detection data set, and then uses res2net to extract the multi receptive field features of the target and proposes a new hybrid feature pyramid structure to improve the network performance from two aspects: fine-grained multi-scale feature extraction and hierarchical multi-scale feature fusion, and realizes a UAV target detection network based on multi-scale feature fusion. Experiments on the UAV detection data set constructed in this article show that the recognition rate of the proposed network to UAV, birds and ordinary aircraft can reach more than 93%.