PyTorchKR Papers

최신 논문

총 7,085편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2410.17251공식 코드 · Papers with Code

Altogether: Image Captioning via Re-aligning Alt-text

This paper focuses on creating synthetic data to improve the quality of image captions. Existing works typically have two shortcomings. First, they caption images from scratch, ignoring existing alt-text metadata, and second, lack transparency if the captioners' training data (e.g. GPT) is unknown. In this paper, we study a principled approach Altogether based on the key idea to edit and re-align existing alt-texts associated with the images. To generate training data, we perform human annotation where annotators start with the existing alt-text and re-align it to the image content in multiple rounds, consequently constructing captions with rich visual concepts. This differs from prior work that carries out human annotation as a one-time description task solely based on images and annotator knowledge. We train a captioner on this data that generalizes the process of re-aligning alt-texts at scale. Our results show our Altogether approach leads to richer image captions that also improve text-to-image generation and zero-shot image classification tasks.

arXiv:2410.16945공식 코드 · Papers with Code

IdenBAT: Disentangled Representation Learning for Identity-Preserved Brain Age Transformation

Brain age transformation aims to convert reference brain images into synthesized images that accurately reflect the age-specific features of a target age group. The primary objective of this task is to modify only the age-related attributes of the reference image while preserving all other age-irrelevant attributes. However, achieving this goal poses substantial challenges due to the inherent entanglement of various image attributes within features extracted from a backbone encoder, resulting in simultaneous alterations during the image generation. To address this challenge, we propose a novel architecture that employs disentangled representation learning for identity-preserved brain age transformation called IdenBAT. This approach facilitates the decomposition of image features, ensuring the preservation of individual traits while selectively transforming age-related characteristics to match those of the target age group. Through comprehensive experiments conducted on both 2D and full-size 3D brain datasets, our method adeptly converts input images to target age while retaining individual characteristics accurately. Furthermore, our approach demonstrates superiority over existing state-of-the-art regarding performance fidelity.

arXiv:2410.16048

Continuous Speech Synthesis using per-token Latent Diffusion

The success of autoregressive transformer models with discrete tokens has inspired quantization-based approaches for continuous modalities, though these often limit reconstruction quality. We therefore introduce SALAD, a per-token latent diffusion model for zero-shot text-to-speech, that operates on continuous representations. SALAD builds upon the recently proposed expressive diffusion head for image generation, and extends it to generate variable-length outputs. Our approach utilizes semantic tokens for providing contextual information and determining the stopping condition. We suggest three continuous variants for our method, extending popular discrete speech synthesis techniques. Additionally, we implement discrete baselines for each variant and conduct a comparative analysis of discrete versus continuous speech modeling techniques. Our results demonstrate that both continuous and discrete approaches are highly competent, and that SALAD achieves a superior intelligibility score while obtaining speech quality and speaker similarity on par with the ground-truth audio.

arXiv:2410.14758

Improving Vector-Quantized Image Modeling with Latent Consistency-Matching Diffusion

By embedding discrete representations into a continuous latent space, we can leverage continuous-space latent diffusion models to handle generative modeling of discrete data. However, despite their initial success, most latent diffusion methods rely on fixed pretrained embeddings, limiting the benefits of joint training with the diffusion model. While jointly learning the embedding (via reconstruction loss) and the latent diffusion model (via score matching loss) could enhance performance, end-to-end training risks embedding collapse, degrading generation quality. To mitigate this issue, we introduce VQ-LCMD, a continuous-space latent diffusion framework within the embedding space that stabilizes training. VQ-LCMD uses a novel training objective combining the joint embedding-diffusion variational lower bound with a consistency-matching (CM) loss, alongside a shifted cosine noise schedule and random dropping strategy. Experiments on several benchmarks show that the proposed VQ-LCMD yields superior results on FFHQ, LSUN Churches, and LSUN Bedrooms compared to discrete-state latent diffusion models. In particular, VQ-LCMD achieves an FID of 6.81 for class-conditional image generation on ImageNet with 50 steps.

arXiv:2410.14749

CFTS-GAN: Continual Few-Shot Teacher Student for Generative Adversarial Networks

Few-shot and continual learning face two well-known challenges in GANs: overfitting and catastrophic forgetting. Learning new tasks results in catastrophic forgetting in deep learning models. In the case of a few-shot setting, the model learns from a very limited number of samples (e.g. 10 samples), which can lead to overfitting and mode collapse. So, this paper proposes a Continual Few-shot Teacher-Student technique for the generative adversarial network (CFTS-GAN) that considers both challenges together. Our CFTS-GAN uses an adapter module as a student to learn a new task without affecting the previous knowledge. To make the student model efficient in learning new tasks, the knowledge from a teacher model is distilled to the student. In addition, the Cross-Domain Correspondence (CDC) loss is used by both teacher and student to promote diversity and to avoid mode collapse. Moreover, an effective strategy of freezing the discriminator is also utilized for enhancing performance. Qualitative and quantitative results demonstrate more diverse image synthesis and produce qualitative samples comparatively good to very stronger state-of-the-art models.

arXiv:2410.14710

G2D2: Gradient-guided Discrete Diffusion for image inverse problem solving

Recent literature has effectively utilized diffusion models trained on continuous variables as priors for solving inverse problems. Notably, discrete diffusion models with discrete latent codes have shown strong performance, particularly in modalities suited for discrete compressed representations, such as image and motion generation. However, their discrete and non-differentiable nature has limited their application to inverse problems formulated in continuous spaces. This paper presents a novel method for addressing linear inverse problems by leveraging image-generation models based on discrete diffusion as priors. We overcome these limitations by approximating the true posterior distribution with a variational distribution constructed from categorical distributions and continuous relaxation techniques. Furthermore, we employ a star-shaped noise process to mitigate the drawbacks of traditional discrete diffusion models with absorbing states, demonstrating that our method performs comparably to continuous diffusion techniques. To the best of our knowledge, this is the first approach to use discrete diffusion model-based priors for solving image inverse problems.

arXiv:2410.14730

On the Relation Between Linear Diffusion and Power Iteration

Recently, diffusion models have gained popularity due to their impressive generative abilities. These models learn the implicit distribution given by the training dataset, and sample new data by transforming random noise through the reverse process, which can be thought of as gradual denoising. In this work, we examine the generation process as a ``correlation machine'', where random noise is repeatedly enhanced in correlation with the implicit given distribution. To this end, we explore the linear case, where the optimal denoiser in the MSE sense is known to be the PCA projection. This enables us to connect the theory of diffusion models to the spiked covariance model, where the dependence of the denoiser on the noise level and the amount of training data can be expressed analytically, in the rank-1 case. In a series of numerical experiments, we extend this result to general low rank data, and show that low frequencies emerge earlier in the generation process, where the denoising basis vectors are more aligned to the true data with a rate depending on their eigenvalues. This model allows us to show that the linear diffusion model converges in mean to the leading eigenvector of the underlying data, similarly to the prevalent power iteration method. Finally, we empirically demonstrate the applicability of our findings beyond the linear case, in the Jacobians of a deep, non-linear denoiser, used in general image generation tasks.

arXiv:2410.14949공식 코드 · Papers with Code

On the Wasserstein Convergence and Straightness of Rectified Flow

Diffusion models have emerged as a powerful tool for image generation and denoising. Typically, generative models learn a trajectory between the starting noise distribution and the target data distribution. Recently Liu et al. (2023b) proposed Rectified Flow (RF), a generative model that aims to learn straight flow trajectories from noise to data using a sequence of convex optimization problems with close ties to optimal transport. If the trajectory is curved, one must use many Euler discretization steps or novel strategies, such as exponential integrators, to achieve a satisfactory generation quality. In contrast, RF has been shown to theoretically straighten the trajectory through successive rectifications, reducing the number of function evaluations (NFEs) while sampling. It has also been shown empirically that RF may improve the straightness in two rectifications if one can solve the underlying optimization problem within a sufficiently small error. In this paper, we make two contributions. First, we provide a theoretical analysis of the Wasserstein distance between the sampling distribution of RF and the target distribution. Our error rate is characterized by the number of discretization steps and a novel formulation of straightness stronger than that in the original work. Secondly, we present general conditions guaranteeing uniqueness and straightness of 1-RF, which is in line with previous empirical findings. As a byproduct of our analysis, we show that, in one dimension, RF started at the standard Gaussian distribution yields the Monge map. Additionally, we also present empirical results on both simulated and real datasets to validate our theoretical findings. The code is available at https://github.com/bansal-vansh/rectified-flow.

arXiv:2410.14987공식 코드 · Papers with Code

SeaS: Few-shot Industrial Anomaly Image Generation with Separation and Sharing Fine-tuning

Current segmentation methods require many training images and precise masks, while insufficient anomaly images hinder their application in industrial scenarios. To address such an issue, we explore producing diverse anomalies and accurate pixel-wise annotations. By observing the real production lines, we find that anomalies vary randomly in shape and appearance, whereas products hold globally consistent patterns with slight local variations. Such a characteristic inspires us to develop a Separation and Sharing Fine-tuning (SeaS) approach using only a few abnormal and some normal images. Firstly, we propose the Unbalanced Abnormal (UA) Text Prompt tailored to industrial anomaly generation, consisting of one product token and several anomaly tokens. Then, for anomaly images, we propose a Decoupled Anomaly Alignment (DA) loss to bind the attributes of the anomalies to different anomaly tokens. Re-blending such attributes may produce never-seen anomalies, achieving a high diversity of anomalies. For normal images, we propose a Normal-image Alignment (NA) loss to learn the products' key features that are used to synthesize products with both global consistency and local variations. The two training processes are separated but conducted on a shared U-Net. Finally, SeaS produces high-fidelity annotations for the generated anomalies by fusing discriminative features of U-Net and high-resolution VAE features. Extensive evaluations on the challenging MVTec AD and MVTec 3D AD dataset demonstrate the effectiveness of our approach. For anomaly image generation, we achieve 1.88 on IS and 0.34 on IC-LPIPS on MVTec AD dataset, 1.95 on IS and 0.30 on IC-LPIPS on MVTec 3D AD dataset. For downstream task, using our generated anomaly image-mask pairs, three common segmentation methods achieve an average 11.17% improvement on IoU on MVTec AD dataset, and a 15.49% enhancement in IoU on MVTec 3D AD dataset.

arXiv:2410.14919공식 코드 · Papers with Code

Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step

Score identity Distillation (SiD) is a data-free method that has achieved SOTA performance in image generation by leveraging only a pretrained diffusion model, without requiring any training data. However, its ultimate performance is constrained by how accurate the pretrained model captures the true data scores at different stages of the diffusion process. In this paper, we introduce SiDA (SiD with Adversarial Loss), which not only enhances generation quality but also improves distillation efficiency by incorporating real images and adversarial loss. SiDA utilizes the encoder from the generator's score network as a discriminator, allowing it to distinguish between real images and those generated by SiD. The adversarial loss is batch-normalized within each GPU and then combined with the original SiD loss. This integration effectively incorporates the average "fakeness" per GPU batch into the pixel-based SiD loss, enabling SiDA to distill a single-step generator. SiDA converges significantly faster than its predecessor when distilled from scratch, and swiftly improves upon the original model's performance during fine-tuning from a pre-distilled SiD generator. This one-step adversarial distillation method establishes new benchmarks in generation performance when distilling EDM diffusion models, achieving FID scores of 1.110 on ImageNet 64x64. When distilling EDM2 models trained on ImageNet 512x512, our SiDA method surpasses even the largest teacher model, EDM2-XXL, which achieved an FID of 1.81 using classifier-free guidance (CFG) and 63 generation steps. In contrast, SiDA achieves FID scores of 2.156 for size XS, 1.669 for S, 1.488 for M, 1.413 for L, 1.379 for XL, and 1.366 for XXL, all without CFG and in a single generation step. These results highlight substantial improvements across all model sizes. Our code is available at https://github.com/mingyuanzhou/SiD/tree/sida.

arXiv:2410.14844

SYNOSIS: Image synthesis pipeline for machine vision in metal surface inspection

The use of machine learning (ML) methods for development of robust and flexible visual inspection system has shown promising. However their performance is highly dependent on the amount and diversity of training data. This is often restricted not only due to costs but also due to a wide variety of defects and product surfaces which occur with varying frequency. As such, one can not guarantee that the acquired dataset contains enough defect and product surface occurrences which are needed to develop a robust model. Using parametric synthetic dataset generation, it is possible to avoid these issues. In this work, we introduce a complete pipeline which describes in detail how to approach image synthesis for surface inspection - from first acquisition, to texture and defect modeling, data generation, comparison to real data and finally use of the synthetic data to train a defect segmentation model. The pipeline is in detail evaluated for milled and sandblasted aluminum surfaces. In addition to providing an in-depth view into each step, discussion of chosen methods, and presentation of ML results, we provide a comprehensive dual dataset containing both real and synthetic images.

arXiv:2410.15432

MedDiff-FM: A Diffusion-based Foundation Model for Versatile Medical Image Applications

Diffusion models have achieved significant success in both the natural image and medical image domains, encompassing a wide range of applications. Previous investigations in medical images have often been constrained to specific anatomical regions, particular applications, and limited datasets, resulting in isolated diffusion models. This paper introduces a diffusion-based foundation model to address a diverse range of medical image tasks, namely MedDiff-FM. MedDiff-FM leverages 3D CT images from multiple publicly available datasets, covering anatomical regions from head to abdomen, to pre-train a diffusion foundation model, and explores the capabilities of the diffusion foundation model across a variety of application scenarios. The diffusion foundation model handles multi-level image processing both at the image-level and patch-level, and utilizes position embedding to establish multi-level spatial relationships as well as anatomical structures and region classes to control certain anatomical regions. MedDiff-FM manages several downstream tasks seamlessly, including image denoising, anomaly detection, and image synthesis. MedDiff-FM is also capable of performing lesion generation and lesion inpainting by rapidly fine-tuning the diffusion foundation model using ControlNet with task-specific conditions. Experimental results demonstrate the effectiveness of MedDiff-FM in addressing diverse downstream medical image tasks.

arXiv:2410.16257공식 코드 · Papers with Code

Elucidating the design space of language models for image generation

The success of autoregressive (AR) language models in text generation has inspired the computer vision community to adopt Large Language Models (LLMs) for image generation. However, considering the essential differences between text and image modalities, the design space of language models for image generation remains underexplored. We observe that image tokens exhibit greater randomness compared to text tokens, which presents challenges when training with token prediction. Nevertheless, AR models demonstrate their potential by effectively learning patterns even from a seemingly suboptimal optimization problem. Our analysis also reveals that while all models successfully grasp the importance of local information in image generation, smaller models struggle to capture the global context. In contrast, larger models showcase improved capabilities in this area, helping to explain the performance gains achieved when scaling up model size. We further elucidate the design space of language models for vision generation, including tokenizer choice, model choice, model scalability, vocabulary design, and sampling strategy through extensive comparative experiments. Our work is the first to analyze the optimization behavior of language models in vision generation, and we believe it can inspire more effective designs when applying LMs to other domains. Finally, our elucidated language model for image generation, termed as ELM, achieves state-of-the-art performance on the ImageNet 256*256 benchmark. The code is available at https://github.com/Pepperlll/LMforImageGeneration.git.

arXiv:2410.14895

Truncated Consistency Models

Consistency models have recently been introduced to accelerate sampling from diffusion models by directly predicting the solution (i.e., data) of the probability flow ODE (PF ODE) from initial noise. However, the training of consistency models requires learning to map all intermediate points along PF ODE trajectories to their corresponding endpoints. This task is much more challenging than the ultimate objective of one-step generation, which only concerns the PF ODE's noise-to-data mapping. We empirically find that this training paradigm limits the one-step generation performance of consistency models. To address this issue, we generalize consistency training to the truncated time range, which allows the model to ignore denoising tasks at earlier time steps and focus its capacity on generation. We propose a new parameterization of the consistency function and a two-stage training procedure that prevents the truncated-time training from collapsing to a trivial solution. Experiments on CIFAR-10 and ImageNet $64\times64$ datasets show that our method achieves better one-step and two-step FIDs than the state-of-the-art consistency models such as iCT-deep, using more than 2$\times$ smaller networks. Project page: https://truncated-cm.github.io/

arXiv:2410.14170공식 코드 · Papers with Code

Personalized Image Generation with Large Multimodal Models

Personalized content filtering, such as recommender systems, has become a critical infrastructure to alleviate information overload. However, these systems merely filter existing content and are constrained by its limited diversity, making it difficult to meet users' varied content needs. To address this limitation, personalized content generation has emerged as a promising direction with broad applications. Nevertheless, most existing research focuses on personalized text generation, with relatively little attention given to personalized image generation. The limited work in personalized image generation faces challenges in accurately capturing users' visual preferences and needs from noisy user-interacted images and complex multimodal instructions. Worse still, there is a lack of supervised data for training personalized image generation models. To overcome the challenges, we propose a Personalized Image Generation Framework named Pigeon, which adopts exceptional large multimodal models with three dedicated modules to capture users' visual preferences and needs from noisy user history and multimodal instructions. To alleviate the data scarcity, we introduce a two-stage preference alignment scheme, comprising masked preference reconstruction and pairwise preference alignment, to align Pigeon with the personalized image generation task. We apply Pigeon to personalized sticker and movie poster generation, where extensive quantitative results and human evaluation highlight its superiority over various generative baselines.

arXiv:2410.13925공식 코드 · Papers with Code

FiTv2: Scalable and Improved Flexible Vision Transformer for Diffusion Model

\textit{Nature is infinitely resolution-free}. In the context of this reality, existing diffusion models, such as Diffusion Transformers, often face challenges when processing image resolutions outside of their trained domain. To address this limitation, we conceptualize images as sequences of tokens with dynamic sizes, rather than traditional methods that perceive images as fixed-resolution grids. This perspective enables a flexible training strategy that seamlessly accommodates various aspect ratios during both training and inference, thus promoting resolution generalization and eliminating biases introduced by image cropping. On this basis, we present the \textbf{Flexible Vision Transformer} (FiT), a transformer architecture specifically designed for generating images with \textit{unrestricted resolutions and aspect ratios}. We further upgrade the FiT to FiTv2 with several innovative designs, includingthe Query-Key vector normalization, the AdaLN-LoRA module, a rectified flow scheduler, and a Logit-Normal sampler. Enhanced by a meticulously adjusted network structure, FiTv2 exhibits $2\times$ convergence speed of FiT. When incorporating advanced training-free extrapolation techniques, FiTv2 demonstrates remarkable adaptability in both resolution extrapolation and diverse resolution generation. Additionally, our exploration of the scalability of the FiTv2 model reveals that larger models exhibit better computational efficiency. Furthermore, we introduce an efficient post-training strategy to adapt a pre-trained model for the high-resolution generation. Comprehensive experiments demonstrate the exceptional performance of FiTv2 across a broad range of resolutions. We have released all the codes and models at \url{https://github.com/whlzy/FiT} to promote the exploration of diffusion transformer models for arbitrary-resolution image generation.

arXiv:2410.14634공식 코드 · Papers with Code

Parallel Backpropagation for Inverse of a Convolution with Application to Normalizing Flows

The inverse of an invertible convolution is an important operation that comes up in Normalizing Flows, Image Deblurring, etc. The naive algorithm for backpropagation of this operation using Gaussian elimination has running time $O(n^3)$ where $n$ is the number of pixels in the image. We give a fast parallel backpropagation algorithm with running time $O(\sqrt{n})$ for a square image and provide a GPU implementation of the same. Inverse of Convolutions are usually used in Normalizing Flows in the sampling pass, making them slow. We propose to use the Inverse of Convolutions in the forward (image to latent vector) pass of the Normalizing flow. Since the sampling pass is the inverse of the forward pass, it will use convolutions only, resulting in efficient sampling times. We use our parallel backpropagation algorithm to optimize the inverse of the convolution layer, resulting in fast training times. We implement this approach in various Normalizing Flow backbones, resulting in our Inverse-Flow models. We benchmark Inverse-Flow on standard datasets and show significantly improved sampling times with similar bits per dimension compared to previous models.

arXiv:2410.14324공식 코드 · Papers with Code

HiCo: Hierarchical Controllable Diffusion Model for Layout-to-image Generation

The task of layout-to-image generation involves synthesizing images based on the captions of objects and their spatial positions. Existing methods still struggle in complex layout generation, where common bad cases include object missing, inconsistent lighting, conflicting view angles, etc. To effectively address these issues, we propose a \textbf{Hi}erarchical \textbf{Co}ntrollable (HiCo) diffusion model for layout-to-image generation, featuring object seperable conditioning branch structure. Our key insight is to achieve spatial disentanglement through hierarchical modeling of layouts. We use a multi branch structure to represent hierarchy and aggregate them in fusion module. To evaluate the performance of multi-objective controllable layout generation in natural scenes, we introduce the HiCo-7K benchmark, derived from the GRIT-20M dataset and manually cleaned. https://github.com/360CVGroup/HiCo_T2I.

arXiv:2410.14159

Assessing Open-world Forgetting in Generative Image Model Customization

Recent advances in diffusion models have significantly enhanced image generation capabilities. However, customizing these models with new classes often leads to unintended consequences that compromise their reliability. We introduce the concept of open-world forgetting to characterize the vast scope of these unintended alterations. Our work presents the first systematic investigation into open-world forgetting in diffusion models, focusing on semantic and appearance drift of representations. Using zero-shot classification, we demonstrate that even minor model adaptations can lead to significant semantic drift affecting areas far beyond newly introduced concepts, with accuracy drops of up to 60% on previously learned concepts. Our analysis of appearance drift reveals substantial changes in texture and color distributions of generated content. To address these issues, we propose a functional regularization strategy that effectively preserves original capabilities while accommodating new concepts. Through extensive experiments across multiple datasets and evaluation metrics, we demonstrate that our approach significantly reduces both semantic and appearance drift. Our study highlights the importance of considering open-world forgetting in future research on model customization and finetuning methods.

arXiv:2410.13924공식 코드 · Papers with Code

ARKit LabelMaker: A New Scale for Indoor 3D Scene Understanding

The performance of neural networks scales with both their size and the amount of data they have been trained on. This is shown in both language and image generation. However, this requires scaling-friendly network architectures as well as large-scale datasets. Even though scaling-friendly architectures like transformers have emerged for 3D vision tasks, the GPT-moment of 3D vision remains distant due to the lack of training data. In this paper, we introduce ARKit LabelMaker, the first large-scale, real-world 3D dataset with dense semantic annotations. Specifically, we complement ARKitScenes dataset with dense semantic annotations that are automatically generated at scale. To this end, we extend LabelMaker, a recent automatic annotation pipeline, to serve the needs of large-scale pre-training. This involves extending the pipeline with cutting-edge segmentation models as well as making it robust to the challenges of large-scale processing. Further, we push forward the state-of-the-art performance on ScanNet and ScanNet200 dataset with prevalent 3D semantic segmentation models, demonstrating the efficacy of our generated dataset.