PyTorchKR Papers

최신 논문

총 7,085편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2406.18583공식 코드 · Papers with Code

Lumina-Next: Making Lumina-T2X Stronger and Faster with Next-DiT

Lumina-T2X is a nascent family of Flow-based Large Diffusion Transformers that establishes a unified framework for transforming noise into various modalities, such as images and videos, conditioned on text instructions. Despite its promising capabilities, Lumina-T2X still encounters challenges including training instability, slow inference, and extrapolation artifacts. In this paper, we present Lumina-Next, an improved version of Lumina-T2X, showcasing stronger generation performance with increased training and inference efficiency. We begin with a comprehensive analysis of the Flag-DiT architecture and identify several suboptimal components, which we address by introducing the Next-DiT architecture with 3D RoPE and sandwich normalizations. To enable better resolution extrapolation, we thoroughly compare different context extrapolation methods applied to text-to-image generation with 3D RoPE, and propose Frequency- and Time-Aware Scaled RoPE tailored for diffusion transformers. Additionally, we introduced a sigmoid time discretization schedule to reduce sampling steps in solving the Flow ODE and the Context Drop method to merge redundant visual tokens for faster network evaluation, effectively boosting the overall sampling speed. Thanks to these improvements, Lumina-Next not only improves the quality and efficiency of basic text-to-image generation but also demonstrates superior resolution extrapolation capabilities and multilingual generation using decoder-based LLMs as the text encoder, all in a zero-shot manner. To further validate Lumina-Next as a versatile generative framework, we instantiate it on diverse tasks including visual recognition, multi-view, audio, music, and point cloud generation, showcasing strong performance across these domains. By releasing all codes and model weights, we aim to advance the development of next-generation generative AI capable of universal modeling.

arXiv:2406.18545

Visual Analysis of Prediction Uncertainty in Neural Networks for Deep Image Synthesis

Ubiquitous applications of Deep neural networks (DNNs) in different artificial intelligence systems have led to their adoption in solving challenging visualization problems in recent years. While sophisticated DNNs offer an impressive generalization, it is imperative to comprehend the quality, confidence, robustness, and uncertainty associated with their prediction. A thorough understanding of these quantities produces actionable insights that help application scientists make informed decisions. Unfortunately, the intrinsic design principles of the DNNs cannot beget prediction uncertainty, necessitating separate formulations for robust uncertainty-aware models for diverse visualization applications. To that end, this contribution demonstrates how the prediction uncertainty and sensitivity of DNNs can be estimated efficiently using various methods and then interactively compared and contrasted for deep image synthesis tasks. Our inspection suggests that uncertainty-aware deep visualization models generate illustrations of informative and superior quality and diversity. Furthermore, prediction uncertainty improves the robustness and interpretability of deep visualization models, making them practical and convenient for various scientific domains that thrive on visual analyses.

arXiv:2404.00234

Grid Diffusion Models for Text-to-Video Generation

Recent advances in the diffusion models have significantly improved text-to-image generation. However, generating videos from text is a more challenging task than generating images from text, due to the much larger dataset and higher computational cost required. Most existing video generation methods use either a 3D U-Net architecture that considers the temporal dimension or autoregressive generation. These methods require large datasets and are limited in terms of computational costs compared to text-to-image generation. To tackle these challenges, we propose a simple but effective novel grid diffusion for text-to-video generation without temporal dimension in architecture and a large text-video paired dataset. We can generate a high-quality video using a fixed amount of GPU memory regardless of the number of frames by representing the video as a grid image. Additionally, since our method reduces the dimensions of the video to the dimensions of the image, various image-based methods can be applied to videos, such as text-guided video manipulation from image manipulation. Our proposed method outperforms the existing methods in both quantitative and qualitative evaluations, demonstrating the suitability of our model for real-world video generation.

논문 ID: 2023.11940

Predictive microstructure image generation using denoising diffusion probabilistic models

The rapid progress in artificial intelligence (AI) based image generation led to groundbreaking achievements, like OpenAI's DALL-E 2, showcasing state-of-the-art generative models in deep learning and computer vision. Recently, the Denoising Diffusion Probabilistic Model (DDPM) has emerged as a strong contender, excelling in generating high-resolution images with complex features similar to those found in real-world images. In this study, we investigate DDPM's potential as both generator and predictor of scanning electron microscope (SEM) images, encompassing both known and unseen microstructural conditions. To rigorously evaluate DDPM, we curated a comprehensive dataset comprising 27 distinct cast-forged AZ80 magnesium alloy components with varied process parameters and microstructure features. Some conditions were held back during training to test DDPM's predictive abilities for unseen scenarios. Our study demonstrates the model's remarkable capacity to capture the inherent physical relationships between process parameters and microstructure features. We scrutinize the synthesized images alongside real-world SEM counterparts, undertaking a comprehensive analysis of various morphological properties. Remarkably, the results show the model's performance, with an average error of 6.36 % ± 0.42 for measured microstructural properties in seen conditions and an equally impressive 6.67 % ± 0.85 for unseen conditions. This study envisions a transformative shift in materials science, as advanced AI predictive models offer new potential to streamline the laborious process of microstructure image generation.

arXiv:2406.18570

It's a Feature, Not a Bug: Measuring Creative Fluidity in Image Generators

With the rise of freely available image generators, AI-generated art has become the centre of a series of heated debates, one of which concerns the concept of human creativity. Can an image generation AI exhibit ``creativity'' of the same type that artists do, and if so, how does that manifest? Our paper attempts to define and empirically measure one facet of creative behavior in AI, by conducting an experiment to quantify the "fluidity of prompt interpretation", or just "fluidity", in a series of selected popular image generators. To study fluidity, we (1) introduce a clear definition for it, (2) create chains of auto-generated prompts and images seeded with an initial "ground-truth: image, (3) measure these chains' breakage points using preexisting visual and semantic metrics, and (4) use both statistical tests and visual explanations to study these chains and determine whether the image generators used to produce them exhibit significant fluidity.

arXiv:2406.19030공식 코드 · Papers with Code

DiffLoss: unleashing diffusion model as constraint for training image restoration network

Image restoration aims to enhance low quality images, producing high quality images that exhibit natural visual characteristics and fine semantic attributes. Recently, the diffusion model has emerged as a powerful technique for image generation, and it has been explicitly employed as a backbone in image restoration tasks, yielding excellent results. However, it suffers from the drawbacks of slow inference speed and large model parameters due to its intrinsic characteristics. In this paper, we introduce a new perspective that implicitly leverages the diffusion model to assist the training of image restoration network, called DiffLoss, which drives the restoration results to be optimized for naturalness and semantic-aware visual effect. To achieve this, we utilize the mode coverage capability of the diffusion model to approximate the distribution of natural images and explore its ability to capture image semantic attributes. On the one hand, we extract intermediate noise to leverage its modeling capability of the distribution of natural images, which serves as a naturalness-oriented optimization space. On the other hand, we utilize the bottleneck features of diffusion model to harness its semantic attributes serving as a constraint on semantic level. By combining these two designs, the overall loss function is able to improve the perceptual quality of image restoration, resulting in visually pleasing and semantically enhanced outcomes. To validate the effectiveness of our method, we conduct experiments on various common image restoration tasks and benchmarks. Extensive experimental results demonstrate that our approach enhances the visual quality and semantic perception of the restoration network.

arXiv:2406.18967공식 코드 · Papers with Code

Structural Attention: Rethinking Transformer for Unpaired Medical Image Synthesis

Unpaired medical image synthesis aims to provide complementary information for an accurate clinical diagnostics, and address challenges in obtaining aligned multi-modal medical scans. Transformer-based models excel in imaging translation tasks thanks to their ability to capture long-range dependencies. Although effective in supervised training settings, their performance falters in unpaired image synthesis, particularly in synthesizing structural details. This paper empirically demonstrates that, lacking strong inductive biases, Transformer can converge to non-optimal solutions in the absence of paired data. To address this, we introduce UNet Structured Transformer (UNest), a novel architecture incorporating structural inductive biases for unpaired medical image synthesis. We leverage the foundational Segment-Anything Model to precisely extract the foreground structure and perform structural attention within the main anatomy. This guides the model to learn key anatomical regions, thus improving structural synthesis under the lack of supervision in unpaired training. Evaluated on two public datasets, spanning three modalities, i.e., MR, CT, and PET, UNest improves recent methods by up to 19.30% across six medical image synthesis tasks. Our code is released at https://github.com/HieuPhan33/MICCAI2024-UNest.

arXiv:2406.18958공식 코드 · Papers with Code

AnyControl: Create Your Artwork with Versatile Control on Text-to-Image Generation

The field of text-to-image (T2I) generation has made significant progress in recent years, largely driven by advancements in diffusion models. Linguistic control enables effective content creation, but struggles with fine-grained control over image generation. This challenge has been explored, to a great extent, by incorporating additional user-supplied spatial conditions, such as depth maps and edge maps, into pre-trained T2I models through extra encoding. However, multi-control image synthesis still faces several challenges. Specifically, current approaches are limited in handling free combinations of diverse input control signals, overlook the complex relationships among multiple spatial conditions, and often fail to maintain semantic alignment with provided textual prompts. This can lead to suboptimal user experiences. To address these challenges, we propose AnyControl, a multi-control image synthesis framework that supports arbitrary combinations of diverse control signals. AnyControl develops a novel Multi-Control Encoder that extracts a unified multi-modal embedding to guide the generation process. This approach enables a holistic understanding of user inputs, and produces high-quality, faithful results under versatile control signals, as demonstrated by extensive quantitative and qualitative evaluations. Our project page is available in https://any-control.github.io.

arXiv:2406.18944공식 코드 · Papers with Code

Rethinking and Defending Protective Perturbation in Personalized Diffusion Models

Personalized diffusion models (PDMs) have become prominent for adapting pretrained text-to-image models to generate images of specific subjects using minimal training data. However, PDMs are susceptible to minor adversarial perturbations, leading to significant degradation when fine-tuned on corrupted datasets. These vulnerabilities are exploited to create protective perturbations that prevent unauthorized image generation. Existing purification methods attempt to mitigate this issue but often over-purify images, resulting in information loss. In this work, we conduct an in-depth analysis of the fine-tuning process of PDMs through the lens of shortcut learning. We hypothesize and empirically demonstrate that adversarial perturbations induce a latent-space misalignment between images and their text prompts in the CLIP embedding space. This misalignment causes the model to erroneously associate noisy patterns with unique identifiers during fine-tuning, resulting in poor generalization. Based on these insights, we propose a systematic defense framework that includes data purification and contrastive decoupling learning. We first employ off-the-shelf image restoration techniques to realign images with their original semantic meanings in latent space. Then, we introduce contrastive decoupling learning with noise tokens to decouple the learning of personalized concepts from spurious noise patterns. Our study not only uncovers fundamental shortcut learning vulnerabilities in PDMs but also provides a comprehensive evaluation framework for developing stronger protection. Our extensive evaluation demonstrates its superiority over existing purification methods and stronger robustness against adaptive perturbation.

arXiv:2406.18790

MUMU: Bootstrapping Multimodal Image Generation from Text-to-Image Data

We train a model to generate images from multimodal prompts of interleaved text and images such as "a <picture of a man> man and his <picture of a dog> dog in an <picture of a cartoon> animated style." We bootstrap a multimodal dataset by extracting semantically meaningful image crops corresponding to words in the image captions of synthetically generated and publicly available text-image data. Our model, MUMU, is composed of a vision-language model encoder with a diffusion decoder and is trained on a single 8xH100 GPU node. Despite being only trained on crops from the same image, MUMU learns to compose inputs from different images into a coherent output. For example, an input of a realistic person and a cartoon will output the same person in the cartoon style, and an input of a standing subject and a scooter will output the subject riding the scooter. As a result, our model generalizes to tasks such as style transfer and character consistency. Our results show the promise of using multimodal models as general purpose controllers for image generation.

arXiv:2406.18524

MultiDiff: Consistent Novel View Synthesis from a Single Image

We introduce MultiDiff, a novel approach for consistent novel view synthesis of scenes from a single RGB image. The task of synthesizing novel views from a single reference image is highly ill-posed by nature, as there exist multiple, plausible explanations for unobserved areas. To address this issue, we incorporate strong priors in form of monocular depth predictors and video-diffusion models. Monocular depth enables us to condition our model on warped reference images for the target views, increasing geometric stability. The video-diffusion prior provides a strong proxy for 3D scenes, allowing the model to learn continuous and pixel-accurate correspondences across generated images. In contrast to approaches relying on autoregressive image generation that are prone to drifts and error accumulation, MultiDiff jointly synthesizes a sequence of frames yielding high-quality and multi-view consistent results -- even for long-term scene generation with large camera movements, while reducing inference time by an order of magnitude. For additional consistency and image quality improvements, we introduce a novel, structured noise distribution. Our experimental results demonstrate that MultiDiff outperforms state-of-the-art methods on the challenging, real-world datasets RealEstate10K and ScanNet. Finally, our model naturally supports multi-view consistent editing without the need for further tuning.

arXiv:2406.18459공식 코드 · Papers with Code

DiffuseHigh: Training-free Progressive High-Resolution Image Synthesis through Structure Guidance

Large-scale generative models, such as text-to-image diffusion models, have garnered widespread attention across diverse domains due to their creative and high-fidelity image generation. Nonetheless, existing large-scale diffusion models are confined to generating images of up to 1K resolution, which is far from meeting the demands of contemporary commercial applications. Directly sampling higher-resolution images often yields results marred by artifacts such as object repetition and distorted shapes. Addressing the aforementioned issues typically necessitates training or fine-tuning models on higher-resolution datasets. However, this poses a formidable challenge due to the difficulty in collecting large-scale high-resolution images and substantial computational resources. While several preceding works have proposed alternatives to bypass the cumbersome training process, they often fail to produce convincing results. In this work, we probe the generative ability of diffusion models at higher resolution beyond their original capability and propose a novel progressive approach that fully utilizes generated low-resolution images to guide the generation of higher-resolution images. Our method obviates the need for additional training or fine-tuning which significantly lowers the burden of computational costs. Extensive experiments and results validate the efficiency and efficacy of our method. Project page: https://yhyun225.github.io/DiffuseHigh/

arXiv:2406.18198

VDG: Vision-Only Dynamic Gaussian for Driving Simulation

Dynamic Gaussian splatting has led to impressive scene reconstruction and image synthesis advances in novel views. Existing methods, however, heavily rely on pre-computed poses and Gaussian initialization by Structure from Motion (SfM) algorithms or expensive sensors. For the first time, this paper addresses this issue by integrating self-supervised VO into our pose-free dynamic Gaussian method (VDG) to boost pose and depth initialization and static-dynamic decomposition. Moreover, VDG can work with only RGB image input and construct dynamic scenes at a faster speed and larger scenes compared with the pose-free dynamic view-synthesis method. We demonstrate the robustness of our approach via extensive quantitative and qualitative experiments. Our results show favorable performance over the state-of-the-art dynamic view synthesis methods. Additional video and source code will be posted on our project page at https://3d-aigc.github.io/VDG.

arXiv:2406.17396

SyncNoise: Geometrically Consistent Noise Prediction for Text-based 3D Scene Editing

Text-based 2D diffusion models have demonstrated impressive capabilities in image generation and editing. Meanwhile, the 2D diffusion models also exhibit substantial potentials for 3D editing tasks. However, how to achieve consistent edits across multiple viewpoints remains a challenge. While the iterative dataset update method is capable of achieving global consistency, it suffers from slow convergence and over-smoothed textures. We propose SyncNoise, a novel geometry-guided multi-view consistent noise editing approach for high-fidelity 3D scene editing. SyncNoise synchronously edits multiple views with 2D diffusion models while enforcing multi-view noise predictions to be geometrically consistent, which ensures global consistency in both semantic structure and low-frequency appearance. To further enhance local consistency in high-frequency details, we set a group of anchor views and propagate them to their neighboring frames through cross-view reprojection. To improve the reliability of multi-view correspondences, we introduce depth supervision during training to enhance the reconstruction of precise geometries. Our method achieves high-quality 3D editing results respecting the textual instructions, especially in scenes with complex textures, by enhancing geometric consistency at the noise and pixel levels.

arXiv:2406.17100공식 코드 · Papers with Code

FaceScore: Benchmarking and Enhancing Face Quality in Human Generation

Diffusion models (DMs) have achieved significant success in generating imaginative images given textual descriptions. However, they are likely to fall short when it comes to real-life scenarios with intricate details. The low-quality, unrealistic human faces in text-to-image generation are one of the most prominent issues, hindering the wide application of DMs in practice. Targeting addressing such an issue, we first assess the face quality of generations from popular pre-trained DMs with the aid of human annotators and then evaluate the alignment between existing metrics with human judgments. Observing that existing metrics can be unsatisfactory for quantifying face quality, we develop a novel metric named FaceScore (FS) by fine-tuning the widely used ImageReward on a dataset of (win, loss) face pairs cheaply crafted by an inpainting pipeline of DMs. Extensive studies reveal FS enjoys a superior alignment with humans. On the other hand, FS opens up the door for enhancing DMs for better face generation. With FS offering image ratings, we can easily perform preference learning algorithms to refine DMs like SDXL. Comprehensive experiments verify the efficacy of our approach for improving face quality. The code is released at https://github.com/OPPO-Mente-Lab/FaceScore.

arXiv:2406.17473

TSynD: Targeted Synthetic Data Generation for Enhanced Medical Image Classification

The usage of medical image data for the training of large-scale machine learning approaches is particularly challenging due to its scarce availability and the costly generation of data annotations, typically requiring the engagement of medical professionals. The rapid development of generative models allows towards tackling this problem by leveraging large amounts of realistic synthetically generated data for the training process. However, randomly choosing synthetic samples, might not be an optimal strategy. In this work, we investigate the targeted generation of synthetic training data, in order to improve the accuracy and robustness of image classification. Therefore, our approach aims to guide the generative model to synthesize data with high epistemic uncertainty, since large measures of epistemic uncertainty indicate underrepresented data points in the training set. During the image generation we feed images reconstructed by an auto encoder into the classifier and compute the mutual information over the class-probability distribution as a measure for uncertainty.We alter the feature space of the autoencoder through an optimization process with the objective of maximizing the classifier uncertainty on the decoded image. By training on such data we improve the performance and robustness against test time data augmentations and adversarial attacks on several classifications tasks.

arXiv:2406.17462

EvolvED: Evolutionary Embeddings to Understand the Generation Process of Diffusion Models

Diffusion models, widely used in image generation, rely on iterative refinement to generate images from noise. Understanding this data evolution is important for model development and interpretability, yet challenging due to its high-dimensional, iterative nature. Prior works often focus on static or instance-level analyses, missing the iterative and holistic aspects of the generative path. While dimensionality reduction can visualize image evolution for few instances, it does preserve the iterative structure. To address these gaps, we introduce EvolvED, a method that presents a holistic view of the iterative generative process in diffusion models. EvolvED goes beyond instance exploration by leveraging predefined research questions to streamline generative space exploration. Tailored prompts aligned with these questions are used to extract intermediate images, preserving iterative context. Targeted feature extractors trace the evolution of key image attribute evolution, addressing the complexity of high-dimensional outputs. Central to EvolvED is a novel evolutionary embedding algorithm that encodes iterative steps while maintaining semantic relations. It enhances the visualization of data evolution by clustering semantically similar elements within each iteration with t-SNE, grouping elements by iteration, and aligning an instance's elements across iterations. We present rectilinear and radial layouts to represent iterations and support exploration. We apply EvolvED to diffusion models like GLIDE and Stable Diffusion, demonstrating its ability to provide valuable insights into the generative process.

arXiv:2406.17343공식 코드 · Papers with Code

Q-DiT: Accurate Post-Training Quantization for Diffusion Transformers

Recent advancements in diffusion models, particularly the architectural transformation from UNet-based models to Diffusion Transformers (DiTs), significantly improve the quality and scalability of image and video generation. However, despite their impressive capabilities, the substantial computational costs of these large-scale models pose significant challenges for real-world deployment. Post-Training Quantization (PTQ) emerges as a promising solution, enabling model compression and accelerated inference for pretrained models, without the costly retraining. However, research on DiT quantization remains sparse, and existing PTQ frameworks, primarily designed for traditional diffusion models, tend to suffer from biased quantization, leading to notable performance degradation. In this work, we identify that DiTs typically exhibit significant spatial variance in both weights and activations, along with temporal variance in activations. To address these issues, we propose Q-DiT, a novel approach that seamlessly integrates two key techniques: automatic quantization granularity allocation to handle the significant variance of weights and activations across input channels, and sample-wise dynamic activation quantization to adaptively capture activation changes across both timesteps and samples. Extensive experiments conducted on ImageNet and VBench demonstrate the effectiveness of the proposed Q-DiT. Specifically, when quantizing DiT-XL/2 to W6A8 on ImageNet ($256 \times 256$), Q-DiT achieves a remarkable reduction in FID by 1.09 compared to the baseline. Under the more challenging W4A8 setting, it maintains high fidelity in image and video generation, establishing a new benchmark for efficient, high-quality quantization in DiTs. Code is available at \href{https://github.com/Juanerx/Q-DiT}{https://github.com/Juanerx/Q-DiT}.

arXiv:2406.17342

Masked Generative Extractor for Synergistic Representation and 3D Generation of Point Clouds

Representation and generative learning, as reconstruction-based methods, have demonstrated their potential for mutual reinforcement across various domains. In the field of point cloud processing, although existing studies have adopted training strategies from generative models to enhance representational capabilities, these methods are limited by their inability to genuinely generate 3D shapes. To explore the benefits of deeply integrating 3D representation learning and generative learning, we propose an innovative framework called \textit{Point-MGE}. Specifically, this framework first utilizes a vector quantized variational autoencoder to reconstruct a neural field representation of 3D shapes, thereby learning discrete semantic features of point patches. Subsequently, we design a sliding masking ratios to smooth the transition from representation learning to generative learning. Moreover, our method demonstrates strong generalization capability in learning high-capacity models, achieving new state-of-the-art performance across multiple downstream tasks. In shape classification, Point-MGE achieved an accuracy of 94.2% (+1.0%) on the ModelNet40 dataset and 92.9% (+5.5%) on the ScanObjectNN dataset. Experimental results also confirmed that Point-MGE can generate high-quality 3D shapes in both unconditional and conditional settings.

arXiv:2406.16476

ResMaster: Mastering High-Resolution Image Generation via Structural and Fine-Grained Guidance

Diffusion models excel at producing high-quality images; however, scaling to higher resolutions, such as 4K, often results in over-smoothed content, structural distortions, and repetitive patterns. To this end, we introduce ResMaster, a novel, training-free method that empowers resolution-limited diffusion models to generate high-quality images beyond resolution restrictions. Specifically, ResMaster leverages a low-resolution reference image created by a pre-trained diffusion model to provide structural and fine-grained guidance for crafting high-resolution images on a patch-by-patch basis. To ensure a coherent global structure, ResMaster meticulously aligns the low-frequency components of high-resolution patches with the low-resolution reference at each denoising step. For fine-grained guidance, tailored image prompts based on the low-resolution reference and enriched textual prompts produced by a vision-language model are incorporated. This approach could significantly mitigate local pattern distortions and improve detail refinement. Extensive experiments validate that ResMaster sets a new benchmark for high-resolution image generation and demonstrates promising efficiency. The project page is https://shuweis.github.io/ResMaster .