Text To Image Generation On Coco

评估指标

FID

Inception score

评测结果

各个模型在此基准测试上的表现结果

			Paper Title	Repository
StackGAN-v1	74.05	8.45	StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks
StackGAN + OP	55.30	12.12	Generating Multiple Objects at Spatially Distinct Locations
L-Verse	45.8	-	L-Verse: Bidirectional Generation Between Image and Text
L-Verse-CC	37.2	-	L-Verse: Bidirectional Generation Between Image and Text
AttnGAN (256 x 256)	35.2	23.3	NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion
AttnGAN + OP	33.35	24.76	Generating Multiple Objects at Spatially Distinct Locations
DM-GAN	32.64	30.49	DM-GAN: Dynamic Memory Generative Adversarial Networks for Text-to-Image Synthesis
DM-GAN + VICTR	32.37	32.37	VICTR: Visual Information Captured Text Representation for Text-to-Image Multimodal Tasks
Vanilla CM3	29.5	-	Retrieval-Augmented Multimodal Language Modeling	-
AttnGAN + VICTR	29.26	28.18	VICTR: Visual Information Captured Text Representation for Text-to-Image Multimodal Tasks
DALL-E (12B)	28	-	Retrieval-Augmented Multimodal Language Modeling	-
DALL-E (256 x 256)	27.5	17.9	NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion
CogView	27.1	18.2	CogView: Mastering Text-to-Image Generation via Transformers
CogView (256 x 256)	27.1	18.2	NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion
Lafite (zero-shot)	26.94	26.02	LAFITE: Towards Language-Free Training for Text-to-Image Generation
DM-GAN (256 x 256)	26.0	32.2	NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion
OP-GAN	24.70	27.88	Semantic Object Accuracy for Generative Text-to-Image Synthesis
CogView2(6B, Finetuned)	24	-	CogView2: Faster and Better Text-to-Image Generation via Hierarchical Transformers
AttnGAN+CL	23.93	25.70	Improving Text-to-Image Synthesis Using Contrastive Learning
FuseDream (k=10, 256)	21.89	34.67	FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimization

0 of 69 row(s) selected.