StylizedGS is a novel 3D neural style transfer framework that leverages 3D Gaussian Splatting (3DGS) to achieve efficient and controllable stylization of 3D scenes. The method addresses the limitations of existing NeRF-based 3D stylization methods, which suffer from efficiency issues and lack accurate geometric pattern style transfer. StylizedGS introduces a filter-based refinement to eliminate floaters in the 3DGS representation, ensuring high-quality stylization. It employs a nearest neighbor-based style loss to fine-tune the geometry and color parameters of 3DGS, while a depth preservation loss and other regularizations prevent geometric content tampering. Users can control color, scale, and spatial regions during the stylization process, enhancing customization capabilities. Extensive experiments demonstrate the effectiveness and efficiency of StylizedGS in terms of stylization quality and inference speed, outperforming state-of-the-art methods in both qualitative and quantitative evaluations.StylizedGS is a novel 3D neural style transfer framework that leverages 3D Gaussian Splatting (3DGS) to achieve efficient and controllable stylization of 3D scenes. The method addresses the limitations of existing NeRF-based 3D stylization methods, which suffer from efficiency issues and lack accurate geometric pattern style transfer. StylizedGS introduces a filter-based refinement to eliminate floaters in the 3DGS representation, ensuring high-quality stylization. It employs a nearest neighbor-based style loss to fine-tune the geometry and color parameters of 3DGS, while a depth preservation loss and other regularizations prevent geometric content tampering. Users can control color, scale, and spatial regions during the stylization process, enhancing customization capabilities. Extensive experiments demonstrate the effectiveness and efficiency of StylizedGS in terms of stylization quality and inference speed, outperforming state-of-the-art methods in both qualitative and quantitative evaluations.