List your company

Text-to-Image Generation on COCO (test)

3.22FIDParti
0.690417.765234.8451.9148Dec 2, 2021May 23, 2022Nov 12, 2022May 3, 2023Oct 23, 2023Apr 12, 2024Oct 2, 2024

Evaluation results

MethodLinks
PartiType=Autoregressive, Pre-training images=4.8B, #Params=20B2024.103.22—————
Ours (Linear Extrapolation)Type=Diffusion, Pre-training images=7M, #Params=464M2024.105—————
Re-ImagenType=Diffusion, Pre-training images=50M, #Params=2.5B2024.105.25—————
Real Images2021.126.0934.8882.8468.58——
Make-A-SceneType=Autoregressive, Pre-training images=35M, #Params=4B2024.107.55—————
VQ-DiffusionType=Diffusion, Pre-training images=15M, #Params=370M2024.1019.75—————
FuseDreamk=5, Resolution=2562021.1221.1634.2696.4365.56——
FuseDreamk=5, Resolution=5122021.1221.5234.1998.3863.67——
FuseDreamk=10, Resolution=2562021.1221.8934.6798.4666.06——
OP-GAN2021.1224.727.88—89.01——
FuseDreamk=10, Resolution=5122021.1225.2432.8898.4463.8——
Obj-GAN2021.1225.8527.37—86.2——
CogView2021.1227.118.2————
DM-GAN2021.1227.3432.3265.4591.87——
DALL-E2021.1227.517.9————
AttnGAN2021.1228.7623.7965.6682.98——
P-AttnGAN2021.1241.5126.31—86.71——
BigSleep2021.1266.4613.3291.8928.75——
AttnGAN2024.10————35.49—
DAE-GAN2024.10————28.12—
DF-GAN2021.12—21.4266.42———
DF-GAN2024.10————21.42—
DM-GAN2024.10————32.64—
FlexTok-1.3BToken structure=1D ordered, Generation paradigm=AR decoding2026.04—————80.4
FlexTok-1.3BToken structure=1D ordered, Generation paradigm=Beam search2026.04—————90
GALIP2024.10————5.85—
GAN-INT-CLS2021.12—7.88————
Infinity-2BToken structure=2D multi-scale, Generation paradigm=AR decoding2026.04—————81.9
Infinity-2BToken structure=2D multi-scale, Generation paradigm=Beam search2026.04—————88.1
Janus-1.3BToken structure=2D grid, Generation paradigm=AR decoding2026.04—————82.3
Janus-1.3BToken structure=2D grid, Generation paradigm=Beam search2026.04—————87.6
Ours2024.10————5—
RAT-GAN2024.10————14.6—
StackGAN2021.12—8.45————
StackGAN++2024.10————81.59—
U-ViT2024.10————5.45—
VQ-Diffusion2024.10————13.86—