Publications

AlignCAT: Visual-Linguistic Alignment of Category and Attribute for Weakly Supervised Visual Grounding

Dynamic Scoring with Enhanced Semantics for Training-Free Human-Object Interaction Detection

FreeInsert: Disentangled Text-Guided Object Insertion in 3D Gaussian Scene without Spatial Priors

Unveiling Open-set Noise: Theoretical Insights into Label Noise

FedMVP: Federated Multimodal Visual Prompt Tuning for Vision-Language Models

Generate, Refine, and Encode: Leveraging Synthesized Novel Samples for On-the-Fly Fine-Grained Category Discovery

Hierarchical Visual Prompt Learning for Continual Video Instance Segmentation

LOTS of Fashion! Multi-Conditioning for Image Generation via Sketch-Text Pairing

On Large Multimodal Models as Open-World Image Classifiers

Pseudo-SD: Pseudo Controlled Stable Diffusion for Semi-Supervised and Cross-Domain Semantic Segmentation

SceneSplat: Gaussian Splatting-based Scene Understanding with Vision-Language Pretraining

Superpowering Open-Vocabulary Object Detectors for X-ray Vision

Training-Free Personalization via Retrieval and Reasoning on Fingerprints

What Changed? Detecting and Evaluating Instruction-Guided Image Edits with Multimodal Large Language Models

Automatic benchmarking of large multimodal models via iterative experiment programming

Diversified in-domain synthesis with efficient fine-tuning for few-shot classification

Evaluating Attribute Confusion in Fashion Text-to-Image Generation

Classifier-to-Bias: Toward Unsupervised Automatic Bias Detection for Visual Classifiers

Compositional Caching for Training-free Open-vocabulary Attribute Detection

Multi-focal Conditioned Latent Diffusion for Person Image Synthesis

Not Only Text: Exploring Compositionality of Visual Representations in Vision-Language Models

Seeing the abstract: Translating the abstract language for vision language models

3D Part Segmentation via Geometric Aggregation of 2D Visual Features

Face Anonymization Made Simple

One vlm to keep it learning: Generation and balancing for data-free continual visual question answering