This study introduces a cutting-edge semantic segmentation framework for LIDAR point cloud data to enhance computer animation and virtual reality applications. Given the inherent challenges of point cloud data such as sparsity, heterogeneity, and multiscale features, our innovative approach incorporates an efficient hybrid attention network focused on significantly improving segmentation accuracy and robustness. The framework comprises an Information Extraction Model (IEM) and an Information Assembly Module (IAM). The IEM utilizes a coordinate-oriented attention mechanism for precise spatial information capture during encoding, effectively handling point cloud sparsity and clutter. During decoding, a depth-separable convolution-based channel attention mechanism optimizes feature channels’ importance, addressing category imbalance issues. The IAM is tasked with the deep fusion of global information for refined feature representation, enhancing semantic segmentation performance. Our hybrid attention mechanism improves adaptability to point cloud data complexity and computational efficiency. Experiments demonstrate our method’s superiority over current state-of-the-art techniques, marking a significant advancement in point cloud processing for realistic virtual environment construction and interaction.