2023-03-30 更新
Text-to-Image Diffusion Models are Zero-Shot Classifiers
Authors:Kevin Clark, Priyank Jaini
The excellent generative capabilities of text-to-image diffusion models suggest they learn informative representations of image-text data. However, what knowledge their representations capture is not fully understood, and they have not been thoroughly explored on downstream tasks. We investigate diffusion models by proposing a method for evaluating them as zero-shot classifiers. The key idea is using a diffusion model’s ability to denoise a noised image given a text description of a label as a proxy for that label’s likelihood. We apply our method to Imagen, using it to probe fine-grained aspects of Imagen’s knowledge and comparing it with CLIP’s zero-shot abilities. Imagen performs competitively with CLIP on a wide range of zero-shot image classification datasets. Additionally, it achieves state-of-the-art results on shape/texture bias tests and can successfully perform attribute binding while CLIP cannot. Although generative pre-training is prevalent in NLP, visual foundation models often use other methods such as contrastive learning. Based on our findings, we argue that generative pre-training should be explored as a compelling alternative for vision and vision-language problems.
PDF
点此查看论文截图
Diffusion Models for Memory-efficient Processing of 3D Medical Images
Authors:Florentin Bieder, Julia Wolleb, Alicia Durrer, Robin Sandkühler, Philippe C. Cattin
Denoising diffusion models have recently achieved state-of-the-art performance in many image-generation tasks. They do, however, require a large amount of computational resources. This limits their application to medical tasks, where we often deal with large 3D volumes, like high-resolution three-dimensional data. In this work, we present a number of different ways to reduce the resource consumption for 3D diffusion models and apply them to a dataset of 3D images. The main contribution of this paper is the memory-efficient patch-based diffusion model \textit{PatchDDM}, which can be applied to the total volume during inference while the training is performed only on patches. While the proposed diffusion model can be applied to any image generation tasks, we evaluate the method on the tumor segmentation task of the BraTS2020 dataset and demonstrate that we can generate meaningful three-dimensional segmentations.
PDF Accepted at MIDL 2023
点此查看论文截图
Training-free Style Transfer Emerges from h-space in Diffusion models
Authors:Jaeseok Jeong, Mingi Kwon, Youngjung Uh
Diffusion models (DMs) synthesize high-quality images in various domains. However, controlling their generative process is still hazy because the intermediate variables in the process are not rigorously studied. Recently, StyleCLIP-like editing of DMs is found in the bottleneck of the U-Net, named $h$-space. In this paper, we discover that DMs inherently have disentangled representations for content and style of the resulting images: $h$-space contains the content and the skip connections convey the style. Furthermore, we introduce a principled way to inject content of one image to another considering progressive nature of the generative process. Briefly, given the original generative process, 1) the feature of the source content should be gradually blended, 2) the blended feature should be normalized to preserve the distribution, 3) the change of skip connections due to content injection should be calibrated. Then, the resulting image has the source content with the style of the original image just like image-to-image translation. Interestingly, injecting contents to styles of unseen domains produces harmonization-like style transfer. To the best of our knowledge, our method introduces the first training-free feed-forward style transfer only with an unconditional pretrained frozen generative network. The code is available at https://curryjung.github.io/DiffStyle/.
PDF
点此查看论文截图
Anti-DreamBooth: Protecting users from personalized text-to-image synthesis
Authors:Thanh Van Le, Hao Phung, Thuan Hoang Nguyen, Quan Dao, Ngoc Tran, Anh Tran
Text-to-image diffusion models are nothing but a revolution, allowing anyone, even without design skills, to create realistic images from simple text inputs. With powerful personalization tools like DreamBooth, they can generate images of a specific person just by learning from his/her few reference images. However, when misused, such a powerful and convenient tool can produce fake news or disturbing content targeting any individual victim, posing a severe negative social impact. In this paper, we explore a defense system called Anti-DreamBooth against such malicious use of DreamBooth. The system aims to add subtle noise perturbation to each user’s image before publishing in order to disrupt the generation quality of any DreamBooth model trained on these perturbed images. We investigate a wide range of algorithms for perturbation optimization and extensively evaluate them on two facial datasets over various text-to-image model versions. Despite the complicated formulation of DreamBooth and Diffusion-based text-to-image models, our methods effectively defend users from the malicious use of those models. Their effectiveness withstands even adverse conditions, such as model or prompt/term mismatching between training and testing. Our code will be available at \href{https://github.com/VinAIResearch/Anti-DreamBooth.git}{https://github.com/VinAIResearch/Anti-DreamBooth.git}.
PDF Project page: https://anti-dreambooth.github.io/
点此查看论文截图
Your Diffusion Model is Secretly a Zero-Shot Classifier
Authors:Alexander C. Li, Mihir Prabhudesai, Shivam Duggal, Ellis Brown, Deepak Pathak
The recent wave of large-scale text-to-image diffusion models has dramatically increased our text-based image generation abilities. These models can generate realistic images for a staggering variety of prompts and exhibit impressive compositional generalization abilities. Almost all use cases thus far have solely focused on sampling; however, diffusion models can also provide conditional density estimates, which are useful for tasks beyond image generation. In this paper, we show that the density estimates from large-scale text-to-image diffusion models like Stable Diffusion can be leveraged to perform zero-shot classification without any additional training. Our generative approach to classification, which we call Diffusion Classifier, attains strong results on a variety of benchmarks and outperforms alternative methods of extracting knowledge from diffusion models. Although a gap remains between generative and discriminative approaches on zero-shot recognition tasks, we find that our diffusion-based approach has stronger multimodal relational reasoning abilities than competing discriminative approaches. Finally, we use Diffusion Classifier to extract standard classifiers from class-conditional diffusion models trained on ImageNet. Even though these models are trained with weak augmentations and no regularization, they approach the performance of SOTA discriminative classifiers. Overall, our results are a step toward using generative over discriminative models for downstream tasks. Results and visualizations at https://diffusion-classifier.github.io/
PDF Website at https://diffusion-classifier.github.io/
点此查看论文截图
Rethinking CycleGAN: Improving Quality of GANs for Unpaired Image-to-Image Translation
Authors:Dmitrii Torbunov, Yi Huang, Huan-Hsin Tseng, Haiwang Yu, Jin Huang, Shinjae Yoo, Meifeng Lin, Brett Viren, Yihui Ren
An unpaired image-to-image (I2I) translation technique seeks to find a mapping between two domains of data in a fully unsupervised manner. While the initial solutions to the I2I problem were provided by the generative adversarial neural networks (GANs), currently, diffusion models (DM) hold the state-of-the-art status on the I2I translation benchmarks in terms of FID. Yet, they suffer from some limitations, such as not using data from the source domain during the training, or maintaining consistency of the source and translated images only via simple pixel-wise errors. This work revisits the classic CycleGAN model and equips it with recent advancements in model architectures and model training procedures. The revised model is shown to significantly outperform other advanced GAN- and DM-based competitors on a variety of benchmarks. In the case of Male2Female translation of CelebA, the model achieves over 40% improvement in FID score compared to the state-of-the-art results. This work also demonstrates the ineffectiveness of the pixel-wise I2I translation faithfulness metrics and suggests their revision. The code and trained models are available at https://github.com/LS4GAN/uvcgan2
PDF
点此查看论文截图
A Unified Single-stage Learning Model for Estimating Fiber Orientation Distribution Functions on Heterogeneous Multi-shell Diffusion-weighted MRI
Authors:Tianyuan Yao, Nancy Newlin, Praitayini Kanakaraj, Vishwesh nath, Leon Y Cai, Karthik Ramadass, Kurt Schilling, Bennett A. Landman, Yuankai Huo
Diffusion-weighted (DW) MRI measures the direction and scale of the local diffusion process in every voxel through its spectrum in q-space, typically acquired in one or more shells. Recent developments in micro-structure imaging and multi-tissue decomposition have sparked renewed attention to the radial b-value dependence of the signal. Applications in tissue classification and micro-architecture estimation, therefore, require a signal representation that extends over the radial as well as angular domain. Multiple approaches have been proposed that can model the non-linear relationship between the DW-MRI signal and biological microstructure. In the past few years, many deep learning-based methods have been developed towards faster inference speed and higher inter-scan consistency compared with traditional model-based methods (e.g., multi-shell multi-tissue constrained spherical deconvolution). However, a multi-stage learning strategy is typically required since the learning process relied on various middle representations, such as simple harmonic oscillator reconstruction (SHORE) representation. In this work, we present a unified dynamic network with a single-stage spherical convolutional neural network, which allows efficient fiber orientation distribution function (fODF) estimation through heterogeneous multi-shell diffusion MRI sequences. We study the Human Connectome Project (HCP) young adults with test-retest scans. From the experimental results, the proposed single-stage method outperforms prior multi-stage approaches in repeated fODF estimation with shell dropoff and single-shell DW-MRI sequences.
PDF
点此查看论文截图
HOLODIFFUSION: Training a 3D Diffusion Model using 2D Images
Authors:Animesh Karnewar, Andrea Vedaldi, David Novotny, Niloy Mitra
Diffusion models have emerged as the best approach for generative modeling of 2D images. Part of their success is due to the possibility of training them on millions if not billions of images with a stable learning objective. However, extending these models to 3D remains difficult for two reasons. First, finding a large quantity of 3D training data is much more complex than for 2D images. Second, while it is conceptually trivial to extend the models to operate on 3D rather than 2D grids, the associated cubic growth in memory and compute complexity makes this infeasible. We address the first challenge by introducing a new diffusion setup that can be trained, end-to-end, with only posed 2D images for supervision; and the second challenge by proposing an image formation model that decouples model memory from spatial memory. We evaluate our method on real-world data, using the CO3D dataset which has not been used to train 3D generative models before. We show that our diffusion models are scalable, train robustly, and are competitive in terms of sample quality and fidelity to existing approaches for 3D generative modeling.
PDF CVPR 2023 conference; project page at: https://holodiffusion.github.io/
点此查看论文截图
4D Facial Expression Diffusion Model
Authors:Kaifeng Zou, Sylvain Faisan, Boyang Yu, Sébastien Valette, Hyewon Seo
Facial expression generation is one of the most challenging and long-sought aspects of character animation, with many interesting applications. The challenging task, traditionally having relied heavily on digital craftspersons, remains yet to be explored. In this paper, we introduce a generative framework for generating 3D facial expression sequences (i.e. 4D faces) that can be conditioned on different inputs to animate an arbitrary 3D face mesh. It is composed of two tasks: (1) Learning the generative model that is trained over a set of 3D landmark sequences, and (2) Generating 3D mesh sequences of an input facial mesh driven by the generated landmark sequences. The generative model is based on a Denoising Diffusion Probabilistic Model (DDPM), which has achieved remarkable success in generative tasks of other domains. While it can be trained unconditionally, its reverse process can still be conditioned by various condition signals. This allows us to efficiently develop several downstream tasks involving various conditional generation, by using expression labels, text, partial sequences, or simply a facial geometry. To obtain the full mesh deformation, we then develop a landmark-guided encoder-decoder to apply the geometrical deformation embedded in landmarks on a given facial mesh. Experiments show that our model has learned to generate realistic, quality expressions solely from the dataset of relatively small size, improving over the state-of-the-art methods. Videos and qualitative comparisons with other methods can be found at https://github.com/ZOUKaifeng/4DFM. Code and models will be made available upon acceptance.
PDF
点此查看论文截图
Physics-Driven Diffusion Models for Impact Sound Synthesis from Videos
Authors:Kun Su, Kaizhi Qian, Eli Shlizerman, Antonio Torralba, Chuang Gan
Modeling sounds emitted from physical object interactions is critical for immersive perceptual experiences in real and virtual worlds. Traditional methods of impact sound synthesis use physics simulation to obtain a set of physics parameters that could represent and synthesize the sound. However, they require fine details of both the object geometries and impact locations, which are rarely available in the real world and can not be applied to synthesize impact sounds from common videos. On the other hand, existing video-driven deep learning-based approaches could only capture the weak correspondence between visual content and impact sounds since they lack of physics knowledge. In this work, we propose a physics-driven diffusion model that can synthesize high-fidelity impact sound for a silent video clip. In addition to the video content, we propose to use additional physics priors to guide the impact sound synthesis procedure. The physics priors include both physics parameters that are directly estimated from noisy real-world impact sound examples without sophisticated setup and learned residual parameters that interpret the sound environment via neural networks. We further implement a novel diffusion model with specific training and inference strategies to combine physics priors and visual information for impact sound synthesis. Experimental results show that our model outperforms several existing systems in generating realistic impact sounds. More importantly, the physics-based representations are fully interpretable and transparent, thus enabling us to perform sound editing flexibly.
PDF CVPR 2023. Project page: https://sukun1045.github.io/video-physics-sound-diffusion/
点此查看论文截图