In recent years, multi-object tracking (MOT) has developed rapidly, but its performance remains unsatisfactory when facing scenarios such as occlusion and similar targets. Meanwhile, most popular tracking-by-detection algorithms in the prevailing paradigm focus on utilizing motion features while neglecting the importance of appearance features, considering appearance information to contribute minimally to tracking performance improvements and being highly complex to process. In this paper, we propose an adaptive fusion method for motion and appearance features that achieves more efficient utilization of both modalities by rationally combining these two types of target characteristics, thereby fully leveraging detector capabilities. Additionally, we introduce an attention-based appearance similarity learning and reconstruction module that enhances target appearance representation through refined processing and optimization, enabling more robust tracking. Combining these two proposed approaches, we present RecreTrack—a simple yet powerful multi-object tracker designed to address challenges in occlusion and target similarity scenarios. Our tracker achieves state-of-the-art performance on MOT17 and MOT20 test sets, with extensive experiments demonstrating the effectiveness of our method in challenging scenarios across both datasets.