An official website of the United States government
Diffusion models are powerful, but they require a lot of time and data to train. We propose Patch Diffusion, a generic patch-wise training framework, to significantly reduce the training time costs while improving data efficiency, which thus helps democratize diffusion model training to broader users. At the core of our innovations is a new conditional score function at the patch level, where the patch location in the original image is included as additional coordinate channels, while the patch size is randomized and diversified throughout training to encode the cross-region dependency at multiple scales. Sampling with our method is as easy as in the original diffusion model. Through Patch Diffusion, we could achieve ≥2× faster training, while maintaining comparable or better generation quality. Patch Diffusion meanwhile improves the performance of diffusion models trained on relatively small datasets, e.g., as few as 5,000 images to train from scratch. We achieve outstanding FID scores in line with state-of-the-art benchmarks: 1.77 on CelebA-64×64, 1.93 on AFHQv2-Wild-64×64, and 2.72 on ImageNet-256×256. We share our code and pre-trained models in GitHub.
more »
« less
Learning Stackable and Skippable LEGO Bricks for Efficient, Reconfigurable, and Variable-Resolution Diffusion Modeling
Diffusion models excel at generating photo-realistic images but come with significant computational costs in both training and sampling. While various techniques address these computational challenges, a less-explored issue is designing an efficient and adaptable network backbone for iterative refinement. Current options like U-Net and Vision Transformer often rely on resource-intensive deep networks and lack the flexibility needed for generating images at variable resolutions or with a smaller network than used in training. This study introduces LEGO bricks, which seamlessly integrate Local-feature Enrichment and Global-content Orchestration. These bricks can be stacked to create a test-time reconfigurable diffusion backbone, allowing selective skipping of bricks to reduce sampling costs and generate higher-resolution images than the training data. LEGO bricks enrich local regions with an MLP and transform them using a Transformer block while maintaining a consistent full-resolution image across all bricks. Experimental results demonstrate that LEGO bricks enhance training efficiency, expedite convergence, and facilitate variable-resolution image generation while maintaining strong generative performance. Moreover, LEGO significantly reduces sampling time compared to other methods, establishing it as a valuable enhancement for diffusion models.
more »
« less
- Award ID(s):
- 2212418
- PAR ID:
- 10536743
- Publisher / Repository:
- International Conference on Learning Representations
- Date Published:
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract. We present 4Diff, a 3D-aware diffusion model addressing the exo-to-ego viewpoint translation task—generating first-person (egocentric) view images from the corresponding third-person (exocentric) images. Building on the diffusion model’s ability to generate photorealistic images, we propose a transformer-based diffusion model that incorporates geometry priors through two mechanisms: (i) egocentric point cloud rasterization and (ii) 3D-aware rotary cross-attention. Egocentric point cloud rasterization converts the input exocentric image into an egocentric layout, which is subsequently used by a diffusion image transformer. As a component of the diffusion transformer’s denoiser block, the 3D-aware rotary cross-attention further incorporates 3D information and semantic features from the source exocentric view. Our 4Diff achieves state-of-the-art results on the challenging and diverse Ego-Exo4D multiview dataset and exhibits robust generalization to novel environments not encountered during training. Our code, processed data, and pretrained models are publicly available at https://klauscc.github.io/4diff.more » « less
-
null (Ed.)Training competitive deep video models is an order of magnitude slower than training their counterpart image models. Slow training causes long research cycles, which hinders progress in video understanding research. Following standard practice for training image models, video model training has used a fixed mini-batch shape: a specific number of clips, frames, and spatial size. However, what is the optimal shape? High resolution models perform well, but train slowly. Low resolution models train faster, but are less accurate. Inspired by multigrid methods in numerical optimization, we propose to use variable mini-batch shapes with different spatial-temporal resolutions that are varied according to a schedule. The different shapes arise from resampling the training data on multiple sampling grids. Training is accelerated by scaling up the mini-batch size and learning rate when shrinking the other dimensions. We empirically demonstrate a general and robust grid schedule that yields a significant out-of-the-box training speedup without a loss in accuracy for different models (I3D, non-local, SlowFast), datasets (Kinetics, Something-Something, Charades), and training settings (with and without pre-training, 128 GPUs or 1 GPU). As an illustrative example, the proposed multigrid method trains a ResNet-50 SlowFast network 4.5 x faster (wall-clock time, same hardware) while also improving accuracy (+ 0.8% absolute) on Kinetics-400 compared to baseline training. Code is available online.more » « less
-
The inference stage of diffusion models involves running a reverse-time diffusion stochastic differential equation, transforming samples from a Gaussian latent distribution into samples from a target distribution on a low-dimensional manifold. The intermediate values can be interpreted as noisy images, with the amount of noise determined by the forward diffusion process noise schedule. Boomerang is an approach for local sampling of image manifolds, which involves adding noise to an input image, moving it closer to the latent space, and mapping it back to the image manifold through a partial reverse diffusion process. Boomerang can be used with any pretrained diffusion model without adjustments to the reverse diffusion process, and we present three applications: constructing privacy-preserving datasets with controllable anonymity, increasing generalization performance with Boomerang for data augmentation, and enhancing resolution with a perceptual image enhancement framework.more » « less
-
Generative models have recently gained increasing attention in image generation and editing tasks. However, they often lack a direct connection to object geometry, which is crucial in sensitive domains such as computational anatomy, biology, and robotics. This paper presents a novel framework for Image Generation informed by Geodesic dynamics (IGG) in deformation spaces. Our IGG model comprises two key components: (i) an efficient autoencoder that explicitly learns the geodesic path of image transformations in the latent space; and (ii) a latent geodesic diffusion model that captures the distribution of latent representations of geodesic deformations conditioned on text instructions. By leveraging geodesic paths, our method ensures smooth, topology-preserving, and interpretable deformations, capturing complex variations in image structures while maintaining geometric consistency. We validate the proposed IGG on plant growth data and brain magnetic resonance imaging (MRI). Experimental results show that IGG outperforms the state-of-the-art image generation/editing models with superior performance in generating realistic, high-quality images with preserved object topology and reduced artifacts. Our code is publicly available at https://github.com/nellie689/IGG.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
