paper-with-me

Self-Supervised Image Classification 벤치마크

Self-Supervised Image Classification on ImageNet

288개 결과 · ⬇ CSV · JSON

Top 1 Accuracy

32.6 46.23 59.85 73.47 87.1 2016-03 2026-09 Colorization (AlexNet) — 32.6 (2016-03-28) Colorization (AlexNet) — 32.6 (2016-03-28) Split-Brain (AlexNet) — 35.4 (2016-11-29) Split-Brain (AlexNet) — 35.4 (2016-11-29) Counting (AlexNet) — 34.3 (2017-08-22) Counting (AlexNet) — 34.3 (2017-08-22) Colorisation (improved) (ResNet-101) — 39.6 (2017-08-25) Colorisation (improved) (ResNet-101) — 39.6 (2017-08-25) Rotation (AlexNet) — 38.7 (2018-03-21) Rotation (AlexNet) — 38.7 (2018-03-21) CPC (ResNet-101 V2) — 48.7 (2018-07-10) CPC (ResNet-101 V2) — 48.7 (2018-07-10) DeepCluster (AlexNet) — 41.0 (2018-07-15) DeepCluster (AlexNet) — 41.0 (2018-07-15) Revisited Rotation (RevNet-50 ×4) — 55.4 (2019-01-25) Revisited Rel.Patch.Loc (ResNet50 ×2) — 51.4 (2019-01-25) Revisited Exemplar (ResNet-50 ×3) — 46.0 (2019-01-25) Revisited Jigsaw (ResNet50 ×2) — 44.6 (2019-01-25) Revisited Rotation (RevNet-50 ×4) — 55.4 (2019-01-25) Revisited Rel.Patch.Loc (ResNet50 ×2) — 51.4 (2019-01-25) Revisited Exemplar (ResNet-50 ×3) — 46.0 (2019-01-25) Revisited Jigsaw (ResNet50 ×2) — 44.6 (2019-01-25) LocalAgg (ResNet-50) — 60.2 (2019-03-29) LocalAgg (ResNet-50) — 60.2 (2019-03-29) CPC v2 (ResNet-161) (arxiv v2) — 71.5 (2019-05-22) CPC v2 (ResNet-50) (arxiv v2) — 63.8 (2019-05-22) CPC v2 (ResNet-161) (arxiv v1) — 61.0 (2019-05-22) CPC v2 (ResNet-161) (arxiv v2) — 71.5 (2019-05-22) CPC v2 (ResNet-50) (arxiv v2) — 63.8 (2019-05-22) CPC v2 (ResNet-161) (arxiv v1) — 61.0 (2019-05-22) AMDIM (large) (arxiv v2) — 68.1 (2019-06-03) AMDIM (small) (arxiv v2) — 63.5 (2019-06-03) AMDIM (arxiv v1) — 60.2 (2019-06-03) AMDIM (large) (arxiv v2) — 68.1 (2019-06-03) AMDIM (small) (arxiv v2) — 63.5 (2019-06-03) AMDIM (arxiv v1) — 60.2 (2019-06-03) CMC (ResNet-50 x2) (arxiv v5) — 70.6 (2019-06-13) CMC (ResNet-50) (arxiv v5) — 66.2 (2019-06-13) CMC (ResNet-101) (arxiv v3) — 65.0 (2019-06-13) CMC (ResNet-101) — 60.1 (2019-06-13) CMC (Alexnet/2) — 42.6 (2019-06-13) CMC (ResNet-50 x2) (arxiv v5) — 70.6 (2019-06-13) CMC (ResNet-50) (arxiv v5) — 66.2 (2019-06-13) CMC (ResNet-101) (arxiv v3) — 65.0 (2019-06-13) CMC (ResNet-101) — 60.1 (2019-06-13) CMC (Alexnet/2) — 42.6 (2019-06-13) BigBiGAN (RevNet-50 ×4, BN+CReLU) — 61.3 (2019-07-04) BigBiGAN (RevNet-50 ×4) — 60.8 (2019-07-04) BigBiGAN (ResNet-50, BN+CReLU) — 56.6 (2019-07-04) BigBiGAN (ResNet-50) — 55.4 (2019-07-04) BigBiGAN (RevNet-50 ×4, BN+CReLU) — 61.3 (2019-07-04) BigBiGAN (RevNet-50 ×4) — 60.8 (2019-07-04) BigBiGAN (ResNet-50, BN+CReLU) — 56.6 (2019-07-04) BigBiGAN (ResNet-50) — 55.4 (2019-07-04) MoCo (ResNet-50 4x) — 68.6 (2019-11-13) MoCo (ResNet-50 2x) — 65.4 (2019-11-13) SeLa (ResNet50) (arxiv 3) — 61.5 (2019-11-13) MoCo (ResNet-50) — 60.6 (2019-11-13) SeLa (ResNet50) — 55.7 (2019-11-13) SeLa (AlexNet) (arxiv v3) — 50.0 (2019-11-13) MoCo (ResNet-50 4x) — 68.6 (2019-11-13) MoCo (ResNet-50 2x) — 65.4 (2019-11-13) SeLa (ResNet50) (arxiv 3) — 61.5 (2019-11-13) MoCo (ResNet-50) — 60.6 (2019-11-13) SeLa (ResNet50) — 55.7 (2019-11-13) SeLa (AlexNet) (arxiv v3) — 50.0 (2019-11-13) PIRL — 63.6 (2019-12-04) PIRL — 63.6 (2019-12-04) SimCLR (ResNet-50 4x) — 76.5 (2020-02-13) SimCLR (ResNet-50 2x) — 74.2 (2020-02-13) SimCLR (ResNet-50) — 69.3 (2020-02-13) SimCLR (ResNet-50 4x) — 76.5 (2020-02-13) SimCLR (ResNet-50 2x) — 74.2 (2020-02-13) SimCLR (ResNet-50) — 69.3 (2020-02-13) MoCo v2 (ResNet-50) — 71.1 (2020-03-09) MoCo v2 (ResNet-50) — 71.1 (2020-03-09) PCL (ResNet-50) — 65.9 (2020-05-11) PCL (ResNet-50) — 65.9 (2020-05-11) InfoMin (ResNeXt-152) — 75.2 (2020-05-20) InfoMin (ResNet-50) — 73.0 (2020-05-20) InfoMin (ResNeXt-152) — 75.2 (2020-05-20) InfoMin (ResNet-50) — 73.0 (2020-05-20) BYOL (ResNet-200 x2) — 79.6 (2020-06-13) BYOL (ResNet-50 x4) — 78.6 (2020-06-13) BYOL (ResNet-50 x2) — 77.4 (2020-06-13) BYOL (ResNet-50) — 74.3 (2020-06-13) BYOL (ResNet-200 x2) — 79.6 (2020-06-13) BYOL (ResNet-50 x4) — 78.6 (2020-06-13) BYOL (ResNet-50 x2) — 77.4 (2020-06-13) BYOL (ResNet-50) — 74.3 (2020-06-13) SimCLRv2 (ResNet-152 x3, SK) — 79.8 (2020-06-17) SwAV (ResNet-50 x5) — 78.5 (2020-06-17) SwAV (ResNet-50 x2) — 77.3 (2020-06-17) SimCLRv2 (ResNet-50 x2) — 75.6 (2020-06-17) SwAV (ResNet-50) — 75.3 (2020-06-17) DeepCluster-v2 (ResNet-50) — 75.2 (2020-06-17) SimCLRv2 (ResNet-50) — 71.7 (2020-06-17) SimCLRv2 (ResNet-152 x3, SK) — 79.8 (2020-06-17) SwAV (ResNet-50 x5) — 78.5 (2020-06-17) SwAV (ResNet-50 x2) — 77.3 (2020-06-17) SimCLRv2 (ResNet-50 x2) — 75.6 (2020-06-17) SwAV (ResNet-50) — 75.3 (2020-06-17) DeepCluster-v2 (ResNet-50) — 75.2 (2020-06-17) SimCLRv2 (ResNet-50) — 71.7 (2020-06-17) iGPT-XL (64x64, 15360 features) — 72.0 (2020-07-17) iGPT-XL (64x64, 3072 features) — 68.7 (2020-07-17) iGPT-L (48x48) — 65.2 (2020-07-17) iGPT-L (32x32) — 60.3 (2020-07-17) iGPT-XL (64x64, 15360 features) — 72.0 (2020-07-17) iGPT-XL (64x64, 3072 features) — 68.7 (2020-07-17) iGPT-L (48x48) — 65.2 (2020-07-17) iGPT-L (32x32) — 60.3 (2020-07-17) ReLIC (ResNet-50) — 74.8 (2020-10-15) ReLIC (ResNet-50) — 74.8 (2020-10-15) SimSiam (ResNet-50) — 71.3 (2020-11-20) SimSiam (ResNet-50) — 71.3 (2020-11-20) FNC (ResNet-50) — 74.4 (2020-11-23) FNC (ResNet-50) — 74.4 (2020-11-23) OBoW (ResNet-50) — 73.8 (2020-12-21) OBoW (ResNet-50) — 73.8 (2020-12-21) HEXA — 75.5 (2020-12-25) HEXA — 75.5 (2020-12-25) SEER — 77.5 (2021-03-02) SEER — 77.5 (2021-03-02) Barlow Twins (ResNet-50) — 73.2 (2021-03-04) Barlow Twins (ResNet-50) — 73.2 (2021-03-04) Self-Classifier (ResNet-50) — 74.2 (2021-03-19) Self-Classifier (ResNet-50) — 74.2 (2021-03-19) MoCo v3 (ViT-BN-L/7) — 81.0 (2021-04-05) MoCo v3 (ViT-BN-H) — 79.1 (2021-04-05) MoCo v3 (ViT-H) — 78.1 (2021-04-05) MoCo v3 (ViT-L) — 77.6 (2021-04-05) MoCo v3 (ViT-B/16) — 76.7 (2021-04-05) MoCo v3 (ViT-BN-L/7) — 81.0 (2021-04-05) MoCo v3 (ViT-BN-H) — 79.1 (2021-04-05) MoCo v3 (ViT-H) — 78.1 (2021-04-05) MoCo v3 (ViT-L) — 77.6 (2021-04-05) MoCo v3 (ViT-B/16) — 76.7 (2021-04-05) Triplet (ResNet-50) — 75.9 (2021-04-18) Triplet (ResNet-50) — 75.9 (2021-04-18) DINO (xcit_medium_24_p8) — 80.3 (2021-04-29) DINO (ViT-B/8) — 80.1 (2021-04-29) DINO (ViT-S/8) — 79.7 (2021-04-29) DINO (ViT-B/16) — 78.2 (2021-04-29) DINO (ViT-S/16) — 77.0 (2021-04-29) NNCLR (ResNet-50, multi-crop) — 75.6 (2021-04-29) DINO (ResNet-50) — 75.3 (2021-04-29) DINO (xcit_medium_24_p8) — 80.3 (2021-04-29) DINO (ViT-B/8) — 80.1 (2021-04-29) DINO (ViT-S/8) — 79.7 (2021-04-29) DINO (ViT-B/16) — 78.2 (2021-04-29) DINO (ViT-S/16) — 77.0 (2021-04-29) NNCLR (ResNet-50, multi-crop) — 75.6 (2021-04-29) DINO (ResNet-50) — 75.3 (2021-04-29) DINO (ResMLP-24) — 72.8 (2021-05-07) DINO (ResMLP-12) — 67.5 (2021-05-07) DINO (ResMLP-24) — 72.8 (2021-05-07) DINO (ResMLP-12) — 67.5 (2021-05-07) MoBY (Swin-T) — 75.0 (2021-05-10) MoBY (DeiT-S) — 72.8 (2021-05-10) MoBY (Swin-T) — 75.0 (2021-05-10) MoBY (DeiT-S) — 72.8 (2021-05-10) VICReg (ResNet50) — 73.2 (2021-05-11) VICReg (ResNet50) — 73.2 (2021-05-11) DnC (ResNet-50) — 75.8 (2021-05-17) DnC (ResNet-50) — 75.8 (2021-05-17) CoKe (ResNet-50) — 76.4 (2021-05-24) CoKe (ResNet-50) — 76.4 (2021-05-24) EsViT (Swin-B) — 81.3 (2021-06-17) EsViT(Swin-S) — 80.8 (2021-06-17) EsViT (Swin-B) — 81.3 (2021-06-17) EsViT(Swin-S) — 80.8 (2021-06-17) ReSSL(ResNet-50) 200ep — 74.7 (2021-07-20) ReSSL(ResNet-50) 200ep — 74.7 (2021-07-20) C-BYOL (ResNet-50 2x, 1000 epochs) — 78.8 (2021-09-27) C-BYOL (ResNet-50, 1000 epochs) — 75.6 (2021-09-27) C-BYOL (ResNet-50 2x, 1000 epochs) — 78.8 (2021-09-27) C-BYOL (ResNet-50, 1000 epochs) — 75.6 (2021-09-27) WCL (ResNet-50) — 74.7 (2021-10-10) WCL (ResNet-50) — 74.7 (2021-10-10) MAE (ViT-H) — 76.6 (2021-11-11) MAE (ViT-L) — 75.8 (2021-11-11) MAE (ViT-B) — 68.0 (2021-11-11) MAE (ViT-H) — 76.6 (2021-11-11) MAE (ViT-L) — 75.8 (2021-11-11) MAE (ViT-B) — 68.0 (2021-11-11) iBOT (ViT-L/16) (IN22k) — 82.3 (2021-11-15) iBOT (ViT-L/16) — 81.3 (2021-11-15) iBOT (ViT-L/16) (IN22k) — 82.3 (2021-11-15) iBOT (ViT-L/16) — 81.3 (2021-11-15) SCE (ResNet-50, multi-crop) — 75.4 (2021-11-29) SCE (ResNet-50, multi-crop) — 75.4 (2021-11-29) MMCL (100 epoch, 256 batch size) — 63.8 (2021-12-21) MMCL (100 epoch, 256 batch size) — 63.8 (2021-12-21) ReLICv2 (ResNet-200 x2) — 80.6 (2022-01-13) ReLICv2 (ResNet200) — 79.8 (2022-01-13) ReLICv2 (ResNet-50 4x) — 79.4 (2022-01-13) ReLICv2 (ResNet152) — 79.3 (2022-01-13) ReLICv2 (ResNet-50 x2) — 79.0 (2022-01-13) ReLICv2 (ResNet101) — 78.7 (2022-01-13) ReLICv2 (ResNet-50) — 77.1 (2022-01-13) ReLICv2 (ResNet-200 x2) — 80.6 (2022-01-13) ReLICv2 (ResNet200) — 79.8 (2022-01-13) ReLICv2 (ResNet-50 4x) — 79.4 (2022-01-13) ReLICv2 (ResNet152) — 79.3 (2022-01-13) ReLICv2 (ResNet-50 x2) — 79.0 (2022-01-13) ReLICv2 (ResNet101) — 78.7 (2022-01-13) ReLICv2 (ResNet-50) — 77.1 (2022-01-13) SEERv2 — 79.8 (2022-02-16) SEERv2 — 79.8 (2022-02-16) ReSSL (ResNet-50 w/ Predictor and Stronger Aug) — 76.3 (2022-03-16) ReSSL (ResNet-50 w/ Predictor) — 76.0 (2022-03-16) ReSSL (ResNet-50 w/ Predictor and Stronger Aug) — 76.3 (2022-03-16) ReSSL (ResNet-50 w/ Predictor) — 76.0 (2022-03-16) Mugs (VIT-L/16) — 82.1 (2022-03-27) CaCo (ResNet-50) — 75.7 (2022-03-27) Mugs (VIT-L/16) — 82.1 (2022-03-27) CaCo (ResNet-50) — 75.7 (2022-03-27) MSN (ViT-L/7) — 80.7 (2022-04-14) MSN (ViT-L/7) — 80.7 (2022-04-14) SMoG (ResNet-50 x4) — 79.0 (2022-07-13) SMoG (ResNet-50 x2) — 78.0 (2022-07-13) SMoG (ResNet-50) — 76.4 (2022-07-13) SMoG (ResNet-50 x4) — 79.0 (2022-07-13) SMoG (ResNet-50 x2) — 78.0 (2022-07-13) SMoG (ResNet-50) — 76.4 (2022-07-13) MR BarTwins (MR BarTwins) — 80.4 (2022-11-15) MR BarTwins (MR BarTwins) — 80.4 (2022-11-15) PercMAE (ViT-B, dVAE) — 79.8 (2022-12-30) PercMAE (ViT-B) — 78.1 (2022-12-30) PercMAE (ViT-B, dVAE) — 79.8 (2022-12-30) PercMAE (ViT-B) — 78.1 (2022-12-30) GroCo (ResNet-50) — 73.9 (2023-01-05) GroCo (ResNet-50) — 73.9 (2023-01-05) MV-MR — 74.5 (2023-03-21) MV-MR — 74.5 (2023-03-21) I-VNE+ (ResNet-50) — 72.1 (2023-04-04) I-VNE+ (ResNet-50) — 72.1 (2023-04-04) Unicom (ViT-B/16) — 79.1 (2023-04-12) Unicom (ViT-B/32) — 75.0 (2023-04-12) Unicom (ViT-B/16) — 79.1 (2023-04-12) Unicom (ViT-B/32) — 75.0 (2023-04-12) DINOv2 (ViT-g/14 @448) — 86.7 (2023-04-14) DINOv2 (ViT-g/14) — 86.5 (2023-04-14) DINOv2 distilled (ViT-L/14) — 86.3 (2023-04-14) DINOv2 distilled (ViT-B/14) — 84.5 (2023-04-14) DINOv2 distilled (ViT-S/14) — 81.1 (2023-04-14) DINOv2 (ViT-g/14 @448) — 86.7 (2023-04-14) DINOv2 (ViT-g/14) — 86.5 (2023-04-14) DINOv2 distilled (ViT-L/14) — 86.3 (2023-04-14) DINOv2 distilled (ViT-B/14) — 84.5 (2023-04-14) DINOv2 distilled (ViT-S/14) — 81.1 (2023-04-14) MAE-CT (ViT-H/16) — 82.2 (2023-04-20) MAE-CT (ViT-L/16 — 81.5 (2023-04-20) MAE-CT (ViT-H/16) — 82.2 (2023-04-20) MAE-CT (ViT-L/16 — 81.5 (2023-04-20) DINOv2+reg (ViT-g/14) — 87.1 (2023-09-28) DINOv2+reg (ViT-g/14) — 87.1 (2023-09-28) PGT (PGT-B w/ Flow) — 80.3 (2023-11-30) PGT (PGT-B w/ Flow) — 80.3 (2023-11-30) MIM-Refiner (D2V2-ViT-H/14) — 84.7 (2024-02-15) MIM-Refiner (MAE-ViT-2B/14) — 84.5 (2024-02-15) MIM-Refiner (MAE-ViT-H/14 — 83.7 (2024-02-15) MIM-Refiner (D2V2-ViT-L/16) — 83.5 (2024-02-15) MIM-Refiner (MAE-ViT-L/16) — 82.8 (2024-02-15) MIM-Refiner (D2V2-ViT-H/14) — 84.7 (2024-02-15) MIM-Refiner (MAE-ViT-2B/14) — 84.5 (2024-02-15) MIM-Refiner (MAE-ViT-H/14 — 83.7 (2024-02-15) MIM-Refiner (D2V2-ViT-L/16) — 83.5 (2024-02-15) MIM-Refiner (MAE-ViT-L/16) — 82.8 (2024-02-15) iBOT-vMF (ViT-B/16) — 80.3 (2024-05-17) DINO-vMF (ViT-B/16) — 78.8 (2024-05-17) DINO-vMF (ViT-S/16) — 77.0 (2024-05-17) iBOT-vMF (ViT-B/16) — 80.3 (2024-05-17) DINO-vMF (ViT-B/16) — 78.8 (2024-05-17) DINO-vMF (ViT-S/16) — 77.0 (2024-05-17) BAM (CAFormer-M36) — 79.3 (2024-08-04) BAM (ViT-B/16) — 78.1 (2024-08-04) BAM (CAFormer-M36) — 79.3 (2024-08-04) BAM (ViT-B/16) — 78.1 (2024-08-04) SynCo (ResNet-50) 800ep — 70.6 (2024-10-03) SynCo (ResNet-50) 200ep — 67.9 (2024-10-03) SynCo (ResNet-50) 800ep — 70.6 (2024-10-03) SynCo (ResNet-50) 200ep — 67.9 (2024-10-03) DiGIT — 80.3 (2024-10-16) DiGIT — 80.3 (2024-10-16) Colorization (AlexNet) — 32.6 (2016-03-28) Split-Brain (AlexNet) — 35.4 (2016-11-29) Colorisation (improved) (ResNet-101) — 39.6 (2017-08-25) CPC (ResNet-101 V2) — 48.7 (2018-07-10) Revisited Rotation (RevNet-50 ×4) — 55.4 (2019-01-25) LocalAgg (ResNet-50) — 60.2 (2019-03-29) CPC v2 (ResNet-161) (arxiv v2) — 71.5 (2019-05-22) SimCLR (ResNet-50 4x) — 76.5 (2020-02-13) BYOL (ResNet-200 x2) — 79.6 (2020-06-13) SimCLRv2 (ResNet-152 x3, SK) — 79.8 (2020-06-17) MoCo v3 (ViT-BN-L/7) — 81.0 (2021-04-05) EsViT (Swin-B) — 81.3 (2021-06-17) iBOT (ViT-L/16) (IN22k) — 82.3 (2021-11-15) DINOv2 (ViT-g/14 @448) — 86.7 (2023-04-14) DINOv2+reg (ViT-g/14) — 87.1 (2023-09-28)
RankModel Top 1 AccuracyTop 5 AccuracyNumber of Params Extra Training Data PaperCodeYear
1 DINOv2+reg (ViT-g/14) 87.1–1100M ✓ Vision Transformers Need Registers rwightman/pytorch-image-models · facebookresearch/dinov2 · locuslab/massive-activations · +3 2023
2 DINOv2 (ViT-g/14 @448) 86.7%–1100M ✓ DINOv2: Learning Robust Visual Features without Supervision huggingface/transformers · facebookresearch/dinov2 · roboflow/rf-detr · +23 2023
3 DINOv2 (ViT-g/14) 86.5%–1100M ✓ DINOv2: Learning Robust Visual Features without Supervision huggingface/transformers · facebookresearch/dinov2 · roboflow/rf-detr · +23 2023
4 DINOv2 distilled (ViT-L/14) 86.3%–307M ✓ DINOv2: Learning Robust Visual Features without Supervision huggingface/transformers · facebookresearch/dinov2 · roboflow/rf-detr · +23 2023
5 MIM-Refiner (D2V2-ViT-H/14) 84.7%–632M MIM-Refiner: A Contrastive Learning Boost from Intermediate Pre-Trained Representations ml-jku/MIM-Refiner · BenediktAlkin/vtab1k-pytorch 2024
6 MIM-Refiner (MAE-ViT-2B/14) 84.5%–1890M MIM-Refiner: A Contrastive Learning Boost from Intermediate Pre-Trained Representations ml-jku/MIM-Refiner · BenediktAlkin/vtab1k-pytorch 2024
6 DINOv2 distilled (ViT-B/14) 84.5%–85M ✓ DINOv2: Learning Robust Visual Features without Supervision huggingface/transformers · facebookresearch/dinov2 · roboflow/rf-detr · +23 2023
8 MIM-Refiner (MAE-ViT-H/14 83.7%–632M MIM-Refiner: A Contrastive Learning Boost from Intermediate Pre-Trained Representations ml-jku/MIM-Refiner · BenediktAlkin/vtab1k-pytorch 2024
9 MIM-Refiner (D2V2-ViT-L/16) 83.5%–307M MIM-Refiner: A Contrastive Learning Boost from Intermediate Pre-Trained Representations ml-jku/MIM-Refiner · BenediktAlkin/vtab1k-pytorch 2024
10 MIM-Refiner (MAE-ViT-L/16) 82.8%–307M MIM-Refiner: A Contrastive Learning Boost from Intermediate Pre-Trained Representations ml-jku/MIM-Refiner · BenediktAlkin/vtab1k-pytorch 2024
11 iBOT (ViT-L/16) (IN22k) 82.3%–307M ✓ iBOT: Image BERT Pre-Training with Online Tokenizer bytedance/ibot · birder/birder 2021
12 MAE-CT (ViT-H/16) 82.2%–632M Contrastive Tuning: A Little Help to Make Masked Autoencoders Forget ml-jku/mae-ct 2023
13 Mugs (VIT-L/16) 82.1%–307M Mugs: A Multi-Granular Self-Supervised Learning Framework sail-sg/mugs 2022
14 MAE-CT (ViT-L/16 81.5%–307M Contrastive Tuning: A Little Help to Make Masked Autoencoders Forget ml-jku/mae-ct 2023
15 EsViT (Swin-B) 81.395.587M Efficient Self-supervised Vision Transformers for Representation Learning microsoft/esvit 2021
15 iBOT (ViT-L/16) 81.3%–307M iBOT: Image BERT Pre-Training with Online Tokenizer bytedance/ibot · birder/birder 2021
17 DINOv2 distilled (ViT-S/14) 81.1%–21M ✓ DINOv2: Learning Robust Visual Features without Supervision huggingface/transformers · facebookresearch/dinov2 · roboflow/rf-detr · +23 2023
18 MoCo v3 (ViT-BN-L/7) 81.0%–304M An Empirical Study of Training Self-Supervised Vision Transformers open-mmlab/mmselfsup · facebookresearch/moco-v3 · Westlake-AI/openmixup · +6 2021
19 EsViT(Swin-S) 80.8–49M Efficient Self-supervised Vision Transformers for Representation Learning microsoft/esvit 2021
20 MSN (ViT-L/7) 80.7%–306M Masked Siamese Networks for Label-Efficient Learning lightly-ai/lightly · facebookresearch/msn 2022
1–20 / 288 다음 → 페이지당 10 20 50 100