Image Generation

Federated Hardware-Prompt Learning (FedHP) enables robust cross-hardware SCI training by aligning inconsistent data distributions using a hardware-conditioned prompter, outperforming existing FL metho…

Trust Sampling enhances guided diffusion by iteratively optimizing constrained generation at each step, improving efficiency and accuracy in image and 3D motion generation.

Consistency Purification boosts certified robustness by efficiently purifying noisy images using a one-step generative model, achieving state-of-the-art results while maintaining semantic alignment.

Consistency Diffusion Bridge Models (CDBMs) dramatically speed up diffusion bridge model sampling by learning a consistency function, achieving up to a 50x speedup with improved sample quality.

ColJailBreak cleverly circumvents AI safety filters by first generating safe images and then subtly injecting unsafe content using image editing.

Classification Diffusion Models (CDMs) revolutionize density ratio estimation by integrating the strengths of diffusion models and classifiers, achieving state-of-the-art image generation and likeliho…

Simple averaging of watermarked images reveals hidden patterns, enabling watermark removal and forgery, thus highlighting the vulnerability of content-agnostic watermarking methods.

Researchers developed a new AI-generated image detection method that overcomes the limitation of existing detectors, achieving superior cross-scene generalization by shuffling image patches and traini…

BrainBits reveals that surprisingly little brain information is needed for high-fidelity image & text reconstruction, highlighting the dominance of generative model priors over neural signal extractio…

BIRD: a novel blind image restoration method jointly optimizes degradation model parameters and the restored image, ensuring realistic outputs via fast diffusion inversion and achieving state-of-the-a…

BLAST matrix learns efficient weight structures for faster deep learning inference, achieving significant compression and performance gains on various models.

BitsFusion achieves 7.9x smaller Stable Diffusion models by quantizing UNet weights to 1.99 bits, surprisingly improving image generation quality!

BI-DiffSR, a novel binarized diffusion model, achieves high-quality image super-resolution with significantly reduced memory and computational costs, outperforming existing methods.

BiDM achieves full 1-bit quantization in diffusion models, significantly improving storage and speed without sacrificing image quality, setting a new state-of-the-art.

O-BELM, a novel diffusion model sampler, achieves mathematically exact inversion with superior sampling quality, offering a new gold standard for diffusion model applications.

Autoregressive image generation is revolutionized by eliminating vector quantization, achieving strong results with increased speed using a novel diffusion procedure.

ZoDiac: a novel image watermarking framework leveraging pre-trained stable diffusion models for robust, invisible watermarks resistant to state-of-the-art attacks.

AsyncDiff accelerates diffusion model inference by 2.8x using asynchronous denoising and model parallelism, maintaining near-perfect image quality.

New framework, MAS, effectively mitigates stereotypes in text-to-image generation by aligning the probability distribution of generated images to stereotype-free distributions.

AsCAN, a novel hybrid architecture, achieves superior efficiency and performance in image recognition and generation by asymmetrically combining convolutional and transformer blocks.