HyperAI超神经

首页算力平台文档资讯论文教程数据集百科 SOTA LLM 模型天梯 GPU 天梯顶会

中文

HyperAI超神经

Text Based Image Editing On Pie Bench

评估指标

Background LPIPS

Background PSNR

CLIPSIM

Structure Distance

评测结果

各个模型在此基准测试上的表现结果

					Paper Title	Repository
FireFlow (Add Q)	239.4	16.49	27.33	70.9	FireFlow: Fast Inversion of Rectified Flow for Image Semantic Editing
DDIM Inversion+Prompt-to-Prompt	208.80	17.87	25.01	69.43	Prompt-to-Prompt Image Editing with Cross Attention Control
DDIM Inversion+Pix2Pix-Zero	172.22	20.44	22.80	61.68	Zero-shot Image-to-Image Translation
Direct Inversion+Pix2Pix-Zero	138.98	21.53	23.31	49.22	Direct Inversion: Boosting Diffusion-based Editing with 3 Lines of Code
FireFlow	123.6	23.03	26.02	27.1	FireFlow: Fast Inversion of Rectified Flow for Image Semantic Editing
DDIM Inversion+Plug-and-Play	113.46	22.28	25.41	28.22	Plug-and-Play Diffusion Features for Text-Driven Image-to-Image Translation
DDIM Inversion+MasaCtrl	106.62	22.17	23.96	28.38	MasaCtrl: Tuning-Free Mutual Self-Attention Control for Consistent Image Synthesis and Editing
Direct Inversion+Plug-and-Play	106.06	22.46	25.41	24.29	Direct Inversion: Boosting Diffusion-based Editing with 3 Lines of Code
Direct Inversion+MasaCtrl	87.94	22.64	24.38	24.70	Direct Inversion: Boosting Diffusion-based Editing with 3 Lines of Code
Negative-Prompt Inversion+Prompt-to-Prompt	69.01	26.21	24.61	16.17	Negative-prompt Inversion: Fast Image Inversion for Editing with Text-guided Diffusion Models	-
StyleDiffusion+Prompt-to-Prompt	66.10	26.05	24.78	11.65	StyleDiffusion: Prompt-Embedding Inversion for Text-Based Editing
Null-Text Inversion+Prompt-to-Prompt	60.67	27.03	24.75	13.44	Null-text Inversion for Editing Real Images using Guided Diffusion Models
Virtual Inversion+Prompt-to-Prompt+LCM	55.85	26.64	24.57	15.61	Inversion-Free Image Editing with Natural Language
Direct Inversion+Prompt-to-Prompt	54.55	27.22	25.02	11.65	Direct Inversion: Boosting Diffusion-based Editing with 3 Lines of Code
Virtual Inversion+Prompt-to-Prompt	47.98	27.52	24.89	14.22	Inversion-Free Image Editing with Natural Language
Virtual Inversion+Unified Attention Control+LCM	47.58	28.51	25.03	13.78	Inversion-Free Image Editing with Natural Language
Virtual Inversion+ViMAEdit	45.67	28.27	25.91	12.65	Vision-guided and Mask-enhanced Adaptive Denoising for Prompt-based Image Editing
KV-Edit	9.92	35.87	25.63	1.98	KV-Edit: Training-Free Image Editing for Precise Background Preservation

0 of 18 row(s) selected.