A lightweight infrared ship detection algorithm CBYOLOv5 combined with Knowledge Distillation (KD) is proposed to solve the problem of large amount of parameters and computation in traditional ship detection algorithms. Lightweight network Ghost module is introduced in YOLOv5s backbone network to realize lightweight detection network. A new neck structure of Asymptotic Feature Pyramid Network (AFPN) is introducedwhich can avoid the large semantic gap of nonadjacent level by fusing two adjacent lowlevel features and gradually fusing to higherlevel features. The VFL function is used to improve the imbalance of positive and negative samples in infrared ship target detection tasksso as to improve the overall performance of the model. FinallyKD is adopted to transfer the “knowledge” in the network of teachers with strong learning ability into the improved network model to improve the accuracy of classification and localization. Experimental results show that in infrared ship datasetin comparison with original algorithm YOLOv5sparameter amount is reduced by 38%and mAP is increased by 3.9 percentage pointswhile the model weight file is only 8.96×106which proves the proposed algorithm is effective and has certain practical value.