Feng Zheng

UniAV combines action localization, sound detection, and audio-visual event localization for better video understanding.

Improving text generation quality by selecting cleaner examples.

LLplace simplifies 3D layout design using natural language input.

ARIO standardizes data to improve robot training and adaptability.

LongVALE provides a new benchmark for understanding long videos through audio-visual data.

A new platform where robots can learn interaction and skills like humans.

New techniques improve anomaly detection in noisy data environments across industries.