Xianming Liu

Cross-ControlNet: Training-Free Fusion of Multiple Conditions for Text-to-Image Generation

Self-Improved Prior for All-in-One Image Restoration

Replacing Multi-Layer Perceptron with Taylor Expansion

Reframing Gaussian Splatting Densification with Complexity-Density Consistency of Primitives

Variation-Bounded Losses for Learning with Noisy Labels

Seeking Global Flat Minima in Federated Domain Generalization via Constrained Adversarial Augmentation

Exploring Source View Capability: Improve Generalizable 3D Reconstruction with Multi-view Context from Source Views

Communication-Efficient Federated Learning via Model Update Distillation