Zijian Yu

Existing detectors fail to maintain critical visual information, but a new two-stage approach achieves unprecedented accuracy in distinguishing real from fake 3D faces.

AIGI detection improves significantly by prioritizing perception over traditional reasoning methods, revealing that fine-grained visual understanding is key to authenticity.

Even state-of-the-art AI-generated image detectors struggle when images are cropped, resized, or compressed, revealing a critical gap in real-world robustness.

Even GPT-4 struggles with long-term preference capture in e-commerce, but a lightweight, jointly-trained LLM agent can beat it.