Shiyi Mu

Shifting from texture-based detection to spatial-logic reasoning, LARAD sets a new standard for robust road anomaly detection in autonomous driving.

DDStereo achieves real-time stereo 3D object detection with open-set generalization, outperforming traditional methods in both speed and accuracy.

Hybrid enrollment strategies can significantly boost the stability and accuracy of short-duration speaker verification systems.

User-defined keyword spotting can now achieve state-of-the-art accuracy and adapt to individual speakers with minimal data and parameter updates, thanks to a novel dual-stage matching and multi-modal enrollment framework.