<p>LiDAR technology is extensively utilized in autonomous driving, playing a pivotal role in high-precision environmental perception. By emitting laser pulses and recording their return time, LiDAR generates detailed 3D point cloud data, which is crucial for object detection and tracking. However, traditional algorithms relying on bounding boxes often struggle with densely packed objects and partial occlusions in dynamic environments. To address these challenges, we propose a novel 3D object detection and tracking framework, termed ECF3DMOT, based on the CenterPoint paradigm. Our framework integrates several key components: voxelization for spatial structuring, a Linear Kernel for LiDAR-based 3D Perception (LKLP) to enhance feature interaction, a Path Augmentation Network (PAN), a Feature Pyramid Network (FPN) for multi-scale representation, and a Deformable Attention Transformer (DATT) for adaptive feature fusion. These modules collectively improve data quality, efficiency, and detection accuracy. Experimental results demonstrate that ECF3DMOT achieves 73.2% NDS and 68.4% mAP on the nuScenes dataset, and 83.7% mAP on the KITTI Moderate set, showing competitive performance compared to existing methods. These results validate the effectiveness of our approach in advancing 3D object detection and tracking for autonomous driving.</p>