12 papers
Diagnose-to-Generate: Eliminating Structural Distortions in Flow Models via Patch-Level Grounded Alignment
Yifan Li, Lishuai Gao, Tong Zhang +8
Flow-based image generators achieve strong semantic synthesis but still produce localized structural distortions, such as malformed hands and limbs. Existing post-training methods…
Demystifying Singular Defects in Large Language Models
Haoqi Wang, Tong Zhang, Mathieu Salzmann
Large transformer models are known to produce high-norm tokens. In vision transformers (ViTs), such tokens have been mathematically modeled through the singular vectors of the line…
Self-Ensembling Gaussian Splatting for Few-Shot Novel View Synthesis
Chen Zhao, Xuan Wang, Tong Zhang +2
3D Gaussian Splatting (3DGS) has demonstrated remarkable effectiveness in novel view synthesis (NVS). However, 3DGS tends to overfit when trained with sparse views, limiting its ge…
Unlocking Comics: The AI4VA Dataset for Visual Understanding
Peter Grönquist, Deblina Bhattacharjee, Bahar Aydemir +4
In the evolving landscape of deep learning, there is a pressing need for more comprehensive datasets capable of training models across multiple modalities. Concurrently, in digital…
Data Augmentation via Latent Diffusion for Saliency Prediction
Bahar Aydemir, Deblina Bhattacharjee, Tong Zhang +2
Saliency prediction models are constrained by the limited diversity and quantity of labeled data. Standard data augmentation techniques such as rotating and cropping alter scene co…
SINDER: Repairing the Singular Defects of DINOv2
Haoqi Wang, Tong Zhang, Mathieu Salzmann
Vision Transformer models trained on large-scale datasets, although effective, often exhibit artifacts in the patch token they extract. While such defects can be alleviated by re-t…
Controlling the Fidelity and Diversity of Deep Generative Models via Pseudo Density
Shuangqi Li, Chen Liu, Tong Zhang +3
We introduce an approach to bias deep generative models, such as GANs and diffusion models, towards generating data with either enhanced fidelity or increased diversity. Our approa…
Coherent and Multi-modality Image Inpainting via Latent Space Optimization
Lingzhi Pan, Tong Zhang, Bingyuan Chen +4
With the advancements in denoising diffusion probabilistic models (DDPMs), image inpainting has significantly evolved from merely filling information based on nearby regions to gen…
Mitigating Object Dependencies: Improving Point Cloud Self-Supervised Learning through Object Exchange
Yanhao Wu, Tong Zhang, Wei Ke +3
In the realm of point cloud scene understanding, particularly in indoor scenes, objects are arranged following human habits, resulting in objects of certain semantics being closely…
OMH: Structured Sparsity via Optimally Matched Hierarchy for Unsupervised Semantic Segmentation
Baran Ozaydin, Tong Zhang, Deblina Bhattacharjee +2
Unsupervised Semantic Segmentation (USS) involves segmenting images without relying on predefined labels, aiming to alleviate the burden of extensive human labeling. Existing metho…
DVMNet++: Rethinking Relative Pose Estimation for Unseen Objects
Chen Zhao, Tong Zhang, Zheng Dang +1
Determining the relative pose of a previously unseen object between two images is pivotal to the success of generalizable object pose estimation. Existing approaches typically pred…
Adaptive Multi-step Refinement Network for Robust Point Cloud Registration
Zhi Chen, Yufan Ren, Tong Zhang +4
Point Cloud Registration (PCR) estimates the relative rigid transformation between two point clouds of the same scene. Despite significant progress with learning-based approaches,…