Abstract: In contour milling, to meet quality requirements and improve processing efficiency, an optimization method for process parameters based on contour curvature features is proposed. Considering the influence of contour curvature on machining quality, milling orthogonal experiments are designed for straight lines, convex arcs, and concave arcs, respectively, to obtain surface roughness data under different contour curvatures. Based on the experimental results, surface roughness prediction models for straight lines milling, convex arcs milling, and concave arcs milling are established using a back propagation (BP) neural network improved by the snow ablation optimizer (SAO) algorithm. By taking surface roughness and material removal rate as optimization objectives, a multi-objective optimization model for contour milling machining parameters is constructed, and the non-dominated sorting genetic algorithm II (NSGA-II) algorithm is applied to solve the model, obtaining milling parameters for different contour milling types that satisfy the specified roughness requirements while maximizing material removal rate. The proposed optimization method is applied to actual contour machining, and the results show that the roughness of all contour regions can meet the quality requirements, and the machining efficiency is improved by 17.1% on average.