Weisi Lin

MANGO-Grasp achieves a remarkable 16.57 percentage point boost in zero-shot grasping performance across diverse robotic hands by leveraging geometry-oriented 3D Gaussian representations.

Achieving state-of-the-art image fusion quality with spiking neural networks while slashing inference energy consumption by an order of magnitude.

Achieving competitive saliency detection accuracy with weak supervision could redefine the landscape of annotation efficiency in computer vision.

Calibrating learning rates based on token reliability can reduce reconstruction errors by over 300% in streaming 3D tasks.

LVLMs falter dramatically in video quality understanding as the duration and complexity of content increase, revealing critical gaps in their perceptual reasoning capabilities.