List your company

Image Reconstruction on CelebA HQ (test)

9.54FID (Reconstruction)SODA
3.553243.964184.375124.7859Mar 11, 2023Sep 6, 2023Mar 4, 2024Aug 31, 2024Feb 26, 2025Aug 25, 2025Feb 21, 2026

Evaluation results

MethodLinks
SODALatent Dim=2048, Backbone=ResNet50x22023.119.54—————18.7885.90.273—
Reg-VQfeature size=16x16, codebook size=1024, resolution=256x2562023.0310.09—————22.05———
Gumbel-VQfeature size=16x16, codebook size=1024, resolution=256x2562023.0312.03—————20.92———
VQ-GANfeature size=16x16, codebook size=1024, resolution=256x2562023.0312.71—————22.44———
TransVQCodebook Size=8912, Code Dim=642026.0213.7————————1
NS-VQCodebook Size=8912, Code Dim=642026.0214.01————————1
SimVQCodebook Size=8912, Code Dim=642026.0214.37————————1
NS-VQCode Dimension=3, Codebook Size=89122026.0216.89————————1
TransVQCode Dimension=3, Codebook Size=89122026.0217.35————————1
NS-VQCodebook Size=1024, Code Dim=642026.0217.44————————1
VQGAN-FCCode Dimension=3, Codebook Size=89122026.0217.57————————0.34
TransVQCodebook Size=1024, Code Dim=642026.0217.6————————1
SimVQCodebook Size=1024, Code Dim=642026.0217.77————————1
TransVQCode Dimension=3, Codebook Size=10242026.0217.82————————1
VQGAN-EMACode Dimension=3, Codebook Size=10242026.0217.86————————1
NS-VQCode Dimension=3, Codebook Size=10242026.0217.86————————1
VQGAN-EMACode Dimension=3, Codebook Size=89122026.0217.93————————1
VQGAN-FCCode Dimension=3, Codebook Size=10242026.0218.48————————1
VQGAN-EMACodebook Size=8912, Code Dim=642026.0219.39————————1
VQGAN-EMACodebook Size=1024, Code Dim=642026.0220.81————————1
SimVQCode Dimension=3, Codebook Size=89122026.0221.69————————1
DALL-E2 (with CLIP)Latent Dim=10242023.1121.91—————9.3431.10.484—
TransVQCodebook Size=64, Code Dim=642026.0223.24————————1
SimVQCode Dimension=3, Codebook Size=10242026.0223.35————————1
SimVQCode Dimension=3, Codebook Size=642026.0223.39————————1
NS-VQCodebook Size=64, Code Dim=642026.0223.39————————1
SimVQCodebook Size=64, Code Dim=642026.0223.4————————1
VQGAN-FCCode Dimension=3, Codebook Size=642026.0224.1————————1
VQVAE2Code Dimension=3, Codebook Size=89122026.0224.24————————0.6
VQGAN-FCCodebook Size=64, Code Dim=642026.0224.41————————0.66
TransVQCode Dimension=3, Codebook Size=642026.0224.42————————1
NS-VQCode Dimension=3, Codebook Size=642026.0224.55————————1
Unconditional Diffusion2023.1125.33—————————
VQGAN-FCCodebook Size=1024, Code Dim=642026.0225.77————————0.03
VQVAE2Codebook Size=8912, Code Dim=642026.0225.87————————0.28
VQGAN-FCCodebook Size=8912, Code Dim=642026.0226————————0
VQ-VAEfeature size=16x16, codebook size=1024, resolution=256x2562023.0328.38—————23.39———
VQVAE2Code Dimension=3, Codebook Size=10242026.0231.17————————0.4
VQVAE2Codebook Size=1024, Code Dim=642026.0236.91————————0.57
VQVAE2Code Dimension=3, Codebook Size=642026.0253.25————————1
VQGAN-EMACode Dimension=3, Codebook Size=642026.0260.86————————1
Vanilla Auto-EncoderLatent Dim=20482023.1165.5—————21.6190.60.327—
β-VAELatent Dim=20482023.1178.65—————19.9488.60.314—
B-TCVAELatent Dim=20482023.1180.33—————20.1889.20.315—
DIP-VAE (I)Latent Dim=20482023.1181.02—————19.8388.40.316—
DIP-VAE (II)Latent Dim=20482023.1182.41—————19.9588.70.307—
VQVAE2Codebook Size=64, Code Dim=642026.0285.86————————0.82
FactorVAELatent Dim=20482023.1187.19—————20.17890.331—
AnnealedVAELatent Dim=20482023.11145.28—————15.9468.60.433—
VQGAN-EMACodebook Size=64, Code Dim=642026.02159.21————————1
DiffAELatent Dim=5122023.11——————15.2868.10.392—
ILO2021.06—36.1540.9843.4647.8948.21————
PnPPrior Type=AWGN denoising2021.06—35.6141.5145.7148.0552.24————
PnPPrior Type=artifact-removal (AR)2021.06—39.1944.248.6651.3253.89————
PULSE2021.06—27.4529.9833.0634.2534.77————
REDPrior Type=AWGN denoising2021.06—35.4641.5945.6548.1352.17————
StyleGAN2, WLatent Dim=5122023.11——————16.7666.20.394—
StyleGAN2, W+Latent Dim=512x142023.11——————21.4281.30.345—
TV2021.06—32.1335.2437.4139.3541.29————
VQGANLatent Dim=256x16x162023.11——————23.2877.30.311—