ICCV2023-Diffusion-Papers

Image/Video Synthesis (2D,3D)

FreeDoM: Training-Free Energy-Guided Conditional Diffusion Model

• Paper: https://arxiv.org/pdf/2306.15667.pdf
• Code: https://github.com/vvictoryuki/FreeDoM

BoxDiff: Text-to-Image Synthesis with Training-Free Box-Constrained Diffusion [Layout-to-Image]

• Paper: https://arxiv.org/abs/2307.10816
• Code: https://github.com/Sierkinhane/BoxDiff

Text2Tex: Text-driven Texture Synthesis via Diffusion Models

• Paper: https://daveredrum.github.io/Text2Tex/static/Text2Tex.pdf
• Code: https://github.com/daveredrum/Text2Tex

Tune-A-Video: One-Shot Tuning of Image Diffusion Models for Text-to-Video Generation

• Paper: https://arxiv.org/abs/2212.11565
• Code: https://github.com/showlab/Tune-A-Video

PODIA-3D: Domain Adaptation of 3D Generative Model Across Large Domain Gap Using Pose-Preserved Text-to-Image Diffusion

• Paper: https://arxiv.org/pdf/2304.01900.pdf

Ablating Concepts in Text-to-Image Diffusion Models

• Paper: https://arxiv.org/abs/2303.13516
• Code: https://github.com/nupurkmr9/concept-ablation

Better Aligning Text-to-Image Models with Human Preference

• Paper: https://arxiv.org/abs/2303.14420
• Code: https://github.com/tgxs002/align_sd

Image Editing

Multimodal Garment Designer: Human-Centric Latent Diffusion Models for Fashion Image Editing

• Paper: https://arxiv.org/abs/2304.02051
• Code: https://github.com/aimagelab/multimodal-garment-designer

Pose Estimation

PoseDiffusion: Solving Pose Estimation via Diffusion-aided Bundle Adjustment

• Paper: https://arxiv.org/pdf/2306.15667.pdf
• Code: https://github.com/facebookresearch/PoseDiffusion

Image Detection

DIRE for Diffusion-Generated Image Detection

• Paper: https://arxiv.org/abs/2303.09295
• Code: https://github.com/ZhendongWang6/DIRE

Low-level Vision

DDFM: Denoising Diffusion Model for Multi-Modality Image Fusion

• Pape: https://arxiv.org/abs/2303.06840
• Code : https://github.com/Zhaozixiang1228/MMIF-DDFM

Human Motion

HumanMAC: Masked Motion Completion for Human Motion Prediction

• Paper: https://arxiv.org/abs/2302.03665
• Code: https://github.com/LinghaoChan/HumanMAC
• Web Page: https://lhchen.top/Human-MAC

BeLFusion: Latent Diffusion for Behavior-Driven Human Motion Prediction

• Paper: https://arxiv.org/abs/2211.14304
• Code: https://github.com/BarqueroGerman/BeLFusion

ReMoDiffuse: Retrieval-Augmented Motion Diffusion Model

• Paper: https://arxiv.org/abs/2304.01116

Quantization

Q-Diffusion: Quantizing Diffusion Models

• Paper: http://arxiv.org/abs/2302.04304
• Code: https://github.com/Xiuyu-Li/q-diffusion

Object Detection / Semantic Segmentation

DiffusionDet is the first work of diffusion model for object detection

• Paper: https://arxiv.org/abs/2211.09788
• Code: https://github.com/ShoufaChen/DiffusionDet

DiffuMask: Synthesizing Images with Pixel-level Annotations for Semantic Segmentation Using Diffusion Models

• Paper: https://arxiv.org/abs/2303.11681
• Code: https://github.com/weijiawu/DiffuMask

Medical Image Analysis

Improving 3D Imaging with Pre-Trained Perpendicular 2D Diffusion Models

• Paper: https://arxiv.org/abs/2303.08440