paper-with-me

Papers

Image Understanding Makes for A Good Tokenizer for Image Generation

2024-11-07 · Luting Wang, Yang Zhao, Zijian Zhang, Jiashi Feng, Si Liu, Bingyi Kang

Abstract Modern image generation (IG) models have been shown to capture rich semantics valuable for image understanding (IU) tasks. However, the potential of IU models to improve IG performance remains uncharted. We address this issue using a token-based IG framework, which relies on effective tokenizers to project images into token sequences. Currently, pixel reconstruction (e.g., VQGAN) dominates the training objective for image tokenizers. In contrast, our approach adopts the feature reconstruction objective, where tokenizers are trained by distilling knowledge from pretrained IU encoders. Comprehensive comparisons indicate that tokenizers with strong IU capabilities achieve superior IG performance across a variety of metrics, datasets, tasks, and proposal networks. Notably, VQ-KD CLIP achieves $4.10$ FID on ImageNet-1k (IN-1k). Visualization suggests that the superiority of VQ-KD can be partly attributed to the rich semantics within the VQ-KD codebook. We further introduce a straightforward pipeline to directly transform IU encoders into tokenizers, demonstrating exceptional effectiveness for IG tasks. These discoveries may energize further exploration into image tokenizer research and inspire the community to reassess the relationship between IU and IG. The code is released at https://github.com/magic-research/vector_quantization.

📄 PDF Abstract BibTeX arXiv:2411.04406

Code (1)

magic-research/vector_quantization 공식 구현 pytorch

Tasks

Image Generation

Methods 이 논문이 사용한 방법론

CLIP Contrastive Language-Image Pre-training (CLIP), consisting of a simplified version of ConVIRT trained from scratch, is an efficient method of image representation learning…

Similar Papers 제목 키워드 기반

What Makes for Good Visual Tokenizers for Large Language Models?

2023-05-20 · Guangzhi Wang, Yixiao Ge, Xiaohan Ding, Mohan Kankanhalli 외

We empirically investigate proper pre-training methods to build good visual tokenizers, making Large Language Models (LLMs) powerful Multimodal Large Language Models (MLLMs). In our benchmark, which is curated to evaluat…

Image CaptioningObject CountingQuestion AnsweringVisual Question Answering

Latent Denoising Makes Good Tokenizers

2025-07-21 · Jiawei Yang, Tianhong Li, Lijie Fan, Yonglong Tian 외 arxiv

Despite their fundamental role, it remains unclear what properties could make tokenizers more effective for generative modeling. We observe that modern generative models share a conceptually similar training objective --…

Image Generation

SemHiTok: A Unified Image Tokenizer via Semantic-Guided Hierarchical Codebook for Multimodal Understanding and Generation

2025-03-09 · Zisheng Chen, Chunwei Wang, Xiuwei Chen, Hang Xu 외

We present SemHiTok, a unified image Tokenizer via Semantic-Guided Hierarchical codebook that provides consistent discrete feature representations for multimodal understanding and generation tasks. Recently, unified mult…

What Makes for Good Tokenizers in Vision Transformer?

2022-12-21 · Shengju Qian, Yi Zhu, Wenbo Li, Mu Li 외

The architecture of transformers, which recently witness booming applications in vision tasks, has pivoted against the widespread convolutional paradigm. Relying on the tokenization process that splits inputs into multip…

Wavelet-Based Image Tokenizer for Vision Transformers

2024-05-28 · Zhenhai Zhu, Radu Soricut

Non-overlapping patch-wise convolution is the default image tokenizer for all state-of-the-art vision Transformer (ViT) models. Even though many ViT variants have been proposed to improve its efficiency and accuracy, lit…

Adversarial Attack