An improved adaptive artificial fish swarm algorithm is proposed based on the artificial fish swarm algorithm for 3D path planning of UAV (Unmanned Aerial Vehicle) in complex sea landscapes. Firstly, a 3D model of the landscape is created using the mathematical formula, and the shortest path is selected as the performance evaluation function to ensure the rationality of the path planning. Secondly, considering the slow convergence speed of the traditional artificial fish swarm algorithm in the early stage and the need for precise search to improve the algorithm accuracy in the later stage, adaptive step size and adaptive field of view range are proposed to update the position of individuals. In order to avoid the algorithm falling into local optimum, the social experience position in the fish school is introduced to update the following behavior. Finally, MATLAB is used to compare with the particle swarm optimization algorithm in several maps, and the simulation results show that the improved artificial fish swarm algorithm has better convergence speed and accuracy in 3D path planning.