YOLO-Master: MOE-Accelerated with Specialized Transformers for Enhanced Real-time Detection
This work addresses computational inefficiency and suboptimal performance in real-time object detection for applications like autonomous driving, representing an incremental improvement over existing YOLO-like methods.
The paper tackles the problem of static dense computation in real-time object detection by proposing YOLO-Master, which uses instance-conditional adaptive computation to dynamically allocate resources based on scene complexity, achieving 42.4% AP with 1.62ms latency on MS COCO, outperforming YOLOv13-N by +0.8% mAP and 17.8% faster inference.
Existing Real-Time Object Detection (RTOD) methods commonly adopt YOLO-like architectures for their favorable trade-off between accuracy and speed. However, these models rely on static dense computation that applies uniform processing to all inputs, misallocating representational capacity and computational resources such as over-allocating on trivial scenes while under-serving complex ones. This mismatch results in both computational redundancy and suboptimal detection performance. To overcome this limitation, we propose YOLO-Master, a novel YOLO-like framework that introduces instance-conditional adaptive computation for RTOD. This is achieved through a Efficient Sparse Mixture-of-Experts (ES-MoE) block that dynamically allocates computational resources to each input according to its scene complexity. At its core, a lightweight dynamic routing network guides expert specialization during training through a diversity enhancing objective, encouraging complementary expertise among experts. Additionally, the routing network adaptively learns to activate only the most relevant experts, thereby improving detection performance while minimizing computational overhead during inference. Comprehensive experiments on five large-scale benchmarks demonstrate the superiority of YOLO-Master. On MS COCO, our model achieves 42.4% AP with 1.62ms latency, outperforming YOLOv13-N by +0.8% mAP and 17.8% faster inference. Notably, the gains are most pronounced on challenging dense scenes, while the model preserves efficiency on typical inputs and maintains real-time inference speed. Code will be available.