Abstract: Leap Motion is widely used in various virtual installations due to its high tracking accuracy and good gesture interaction.The Leap Motion gesture recognition is applied to high-precision virtual installation of highly integrated large-scale structural parts,which can realize the interactive control of the installation process by the virtual hand. A fuzzy k-nearest neighbor (KNN) classification method based on weighted chi-square distance is designed to achieve virtual gesture classification. Different weights are assigned according to the importance of gesture features, which can further improve the classification accuracy. The test results show that the improved classification method has a recognition accuracy of 92.7%. It is 5.3% higher than traditional classification algorithms. Three kinds of gestures are used for virtual installation experiments of engine components. The results show that gesture recognition has achieved good results in the installation process, which can improve the quality and efficiency of the actual installation process, and it is of great significance to improve the manufacturing and installation level of large military products.