Browse › Computer Vision › Action Classification › Kinetics-400
Action Classification archive 2025-07-28
Kinetics-400 Benchmark (Action Classification)
Image source: The Kinetics Human Action Video Dataset
The archive carries no text for this table; the description above is the archive's text for the task Action Classification. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Direction inferred from the metric name, not from the archive: Acc@1 (higher is better), Acc@5 (higher is better), FLOPs (G) x views (lower is better), Clip acc@1 (higher is better), Clip acc@5 (higher is better). Not inferred (points only, no best-so-far line): Parameters (M). Points are placed at the row's paper date; 207 of 207 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | OmniVec2 | 93.6 | – | Paper | – | 2024 | no code linked | report | |||||
| 2 | FTP-UniFormerV2-L/14 | 93.4 | – | Paper | – | 2024 | no code linked | report | |||||
| 3 | InternVideo2-6B | 92.1 | ✓ | Paper | Code | 2024 | linked, not harvested | report | |||||
| 4 | InternVideo2-1B | 91.6 | ✓ | Paper | Code | 2024 | linked, not harvested | report | |||||
| 5 | InternVideo | 91.1 | – | Paper | Code | 2022 | 3 of 3 ran · 0 unverified | report | |||||
| 6 | OmniVec | 91.1 | – | Paper | – | 2023 | no code linked | report | |||||
| 7 | TubeViT-H (ImageNet-1k) | 90.9 | 98.9 | 176400x4x3 | 632 | – | Paper | Code | 2022 | 1 of 3 ran · 2 unverified | report | ||
| 8 | Unmasked Teacher (ViT-L) | 90.6 | 98.7 | 1434×3×4 | 304 | – | Paper | Code | 2023 | 3 of 8 ran · 5 unverified | report | ||
| 9 | UMT-L (ViT-L/16) | 90.6 | 98.7 | – | Paper | Code | 2023 | 3 of 8 ran · 5 unverified | report | ||||
| 10 | TubeVit-L (ImageNet-1k) | 90.2 | 98.6 | 95300x4x3 | 307 | – | Paper | Code | 2022 | 1 of 3 ran · 2 unverified | report | ||
| 11 | UniFormerV2-L (ViT-L, 336) | 90.0 | 98.4 | 75300x3x2 | 354 | ✓ | Paper | Code | 2022 | linked, not harvested | report | ||
| 12 | VideoMAE V2-g (64x266x266) | 90.0 | 98.4 | – | Paper | Code | 2023 | 2 of 6 ran · 4 unverified | report | ||||
| 13 | FluxViT-B | 90.0 | 440x3x4 | 97 | ✓ | Paper | Code | 2025 | 7 of 17 ran · 10 unverified | report | |||
| 14 | MTV-H (WTS 60M) | 89.9 | 98.3 | 735700x4x3 | – | Paper | Code | 2022 | linked, not harvested | report | |||
| 15 | TAdaFormer-L/14 | 89.9 | – | Paper | Code | 2023 | 2 of 2 ran · 0 unverified | report | |||||
| 16 | EVA | 89.7 | – | Paper | Code | 2022 | 1 of 3 ran · 2 unverified | report | |||||
| 17 | AM/12 ViT-B Dinov2 | 89.6 | – | Paper | – | 2024 | no code linked | report | |||||
| 18 | ATM | 89.4 | 98.3 | – | Paper | Code | 2023 | 3 of 6 ran · 3 unverified | report | ||||
| 19 | DejaVid | 89.1 | 98.2 | ✓ | Paper | Code | 2025 | linked, not harvested | report | ||||
| 20 | CoCa (finetuned) | 88.9 | – | Paper | Code | 2022 | 9 of 17 ran · 8 unverified | report | |||||
| 21 | BIKE (CLIP ViT-L/14) | 88.7 | 98.4 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 22 | ILA (ViT-L/14) | 88.7 | 97.8 | – | Paper | Code | 2023 | 2 of 4 ran · 2 unverified | report | ||||
| 23 | Side4Video (EVA, ViT-E/14) | 88.6 | 98.2 | – | Paper | Code | 2023 | linked, not harvested | report | ||||
| 24 | TubeVit-B (ImageNet-1k) | 88.6 | 97.6 | 8700x3x4 | 86 | – | Paper | Code | 2022 | 1 of 3 ran · 2 unverified | report | ||
| 25 | VideoMAE V2-g | 88.5 | 98.1 | ✓ | Paper | Code | 2023 | 2 of 6 ran · 4 unverified | report | ||||
| 26 | ONE-PEACE | 88.1 | 97.8 | – | Paper | Code | 2023 | 2 of 7 ran · 5 unverified | report | ||||
| 27 | FluxViT-S | 88.0 | 154x3x4 | 24 | ✓ | Paper | Code | 2025 | 7 of 17 ran · 10 unverified | report | |||
| 28 | CoCa (frozen) | 88.0 | – | Paper | Code | 2022 | 9 of 17 ran · 8 unverified | report | |||||
| 29 | ViT-22B | 88.0 | – | Paper | Code | 2023 | linked, not harvested | report | |||||
| 30 | Text4Vis (CLIP ViT-L/14) | 87.8 | 97.6 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 31 | Hiera-H (no extra data) | 87.8 | – | Paper | Code | 2023 | 0 of 6 ran · 6 unverified | report | |||||
| 32 | EVL (CLIP ViT-L/14@336px, frozen, 32 frames) | 87.7 | 97.8 | – | Paper | Code | 2022 | 5 of 10 ran · 5 unverified | report | ||||
| 33 | DualPath w/ ViT-L/14 | 87.7 | 97.8 | – | Paper | Code | 2023 | linked, not harvested | report | ||||
| 34 | X-CLIP(ViT-L/14, CLIP) | 87.7 | 97.4 | – | Paper | Code | 2022 | 3 of 3 ran · 0 unverified | report | ||||
| 35 | AIM (CLIP ViT-L/14, 32x224) | 87.5 | 97.7 | ✓ | Paper | Code | 2023 | linked, not harvested | report | ||||
| 36 | VideoMAE (no extra data, ViT-H, 32x320x320) | 87.4 | 97.6 | – | Paper | Code | 2022 | 9 of 13 ran · 4 unverified | report | ||||
| 37 | ST-Adapter (ViT-L, CLIP) | 87.2 | 97.6 | – | Paper | Code | 2022 | 4 of 10 ran · 6 unverified | report | ||||
| 38 | ZeroI2V ViT-L/14 | 87.2 | 97.6 | – | Paper | Code | 2023 | 1 of 3 ran · 2 unverified | report | ||||
| 39 | CoVeR (JFT-3B) | 87.2 | 97.5 | – | Paper | – | 2021 | no code linked | report | ||||
| 40 | MVD (K400 pretrain, ViT-H, 16x224x224) | 87.2 | 97.4 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 41 | mPLUG-2 | 87.1 | 97.7 | – | Paper | Code | 2023 | 9 of 19 ran · 10 unverified | report | ||||
| 42 | MaskFeat (K600, MViT-L) | 87.0 | 97.4 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 43 | VicTR (ViT-L/14) | 87.0 | – | Paper | – | 2023 | no code linked | report | |||||
| 44 | Video-SwinV2-G (ImageNet-22k and external 70M pretrain) | 86.8 | – | Paper | Code | 2021 | 3 of 30 ran · 27 unverified | report | |||||
| 45 | MaskFeat (no extra data, MViT-L) | 86.7 | 97.3 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 46 | VideoMAE (no extra data, ViT-H) | 86.6 | 97.1 | – | Paper | Code | 2022 | 9 of 13 ran · 4 unverified | report | ||||
| 47 | MVD (K400 pretrain, ViT-L, 16x224x224) | 86.4 | 97.0 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 48 | TAdaConvNeXtV2-B | 86.4 | – | Paper | Code | 2023 | 2 of 2 ran · 0 unverified | report | |||||
| 49 | CoVeR (JFT-300M) | 86.3 | 97.2 | – | Paper | – | 2021 | no code linked | report | ||||
| 50 | VideoMAE (no extra data, ViT-L, 32x320x320) | 86.1 | 97.3 | – | Paper | Code | 2022 | 9 of 13 ran · 4 unverified | report | ||||
| 51 | MViTv2-L (ImageNet-21k pretrain) | 86.1 | 97.0 | ✓ | Paper | Code | 2021 | linked, not harvested | report | ||||
| 52 | ILA (ViT-B/16) | 85.7 | 97.2 | – | Paper | Code | 2023 | 2 of 4 ran · 2 unverified | report | ||||
| 53 | DualPath w/ ViT-B/16 | 85.4 | 97.1 | – | Paper | Code | 2023 | linked, not harvested | report | ||||
| 54 | TokenLearner 16at18 (L/10) | 85.4 | – | Paper | Code | 2021 | 3 of 3 ran · 0 unverified | report | |||||
| 55 | MAR (50% mask, ViT-L, 16x4) | 85.3 | 96.3 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 56 | CAST(ViT-B/16) | 85.3 | – | Paper | Code | 2023 | 9 of 17 ran · 8 unverified | report | |||||
| 57 | VideoMAE (no extra data, ViT-L, 16x4) | 85.2 | 96.8 | – | Paper | Code | 2022 | 9 of 13 ran · 4 unverified | report | ||||
| 58 | ViC-MAE (ViT-L) | 85.1 | – | Paper | Code | 2023 | linked, not harvested | report | |||||
| 59 | VideoMamba-M800 | 85.0 | – | Paper | Code | 2024 | 9 of 15 ran · 6 unverified | report | |||||
| 60 | Swin-L (384x384, ImageNet-21k pretrain) | 84.9 | 96.7 | – | Paper | Code | 2021 | 7 of 32 ran · 25 unverified | report | ||||
| 61 | ViViT-H/16x2 (JFT) | 84.9 | 95.8 | – | Paper | Code | 2021 | 14 of 21 ran · 7 unverified | report | ||||
| 62 | OMNIVORE (Swin-L) | 84.1 | 96.1 | – | Paper | Code | 2022 | 2 of 2 ran · 0 unverified | report | ||||
| 63 | OMNIVORE (Swin-B) | 84.0 | 96.2 | – | Paper | Code | 2022 | 2 of 2 ran · 0 unverified | report | ||||
| 64 | MAR (75% mask, ViT-L, 16x4) | 83.9 | 96.0 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 65 | ActionCLIP (CLIP-pretrained) | 83.8 | 97.1 | – | Paper | Code | 2021 | 5 of 9 ran · 4 unverified | report | ||||
| 66 | OmniSource irCSN-152 (IG-Kinetics-65M pretrain) | 83.6 | – | Paper | Code | 2020 | 0 of 13 ran · 13 unverified | report | |||||
| 67 | MVD (K400 pretrain, ViT-B, 16x224x224) | 83.4 | 95.8 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 68 | StructViT-B-4-1 | 83.4 | – | Paper | – | 2024 | no code linked | report | |||||
| 69 | Swin-L (ImageNet-21k pretrain) | 83.1 | 95.9 | – | Paper | Code | 2021 | 7 of 32 ran · 25 unverified | report | ||||
| 70 | SIFA | 83.1 | – | Paper | Code | 2022 | linked, not harvested | report | |||||
| 71 | UniFormer-B (ImageNet-1K) | 82.9 | 94.5 | 259x4 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 72 | irCSN-152 (IG-Kinetics-65M pretrain) | 82.8 | – | Paper | Code | 2019 | 0 of 21 ran · 21 unverified | report | |||||
| 73 | DirecFormer | 82.75 | 94.86 | – | Paper | Code | 2022 | 4 of 8 ran · 4 unverified | report | ||||
| 74 | Swin-B (ImageNet-21k pretrain) | 82.7 | 95.5 | – | Paper | Code | 2021 | 7 of 32 ran · 25 unverified | report | ||||
| 75 | ir-CSN-152 (IG-65M pretraining) | 82.6 | – | Paper | Code | 2019 | 1 of 4 ran · 3 unverified | report | |||||
| 76 | ip-CSN-152 (IG-65M pretraining) | 82.5 | 95.3 | – | Paper | Code | 2019 | 1 of 4 ran · 3 unverified | report | ||||
| 77 | TPS | 82.5 | – | Paper | Code | 2022 | 1 of 1 ran · 0 unverified | report | |||||
| 78 | ILA (ViT-B/32) | 82.4 | 95.8 | – | Paper | Code | 2023 | 2 of 4 ran · 2 unverified | report | ||||
| 79 | AMD(ViT-B/16) | 82.2 | 95.3 | 180x15 | 87 | – | Paper | – | 2023 | no code linked | report | ||
| 80 | VATT-Large | 82.1 | 95.5 | – | Paper | Code | 2021 | 5 of 8 ran · 3 unverified | report | ||||
| 81 | AdaMAE | 81.7 | 95.2 | – | Paper | Code | 2022 | 9 of 25 ran · 16 unverified | report | ||||
| 82 | VideoMAE (no extra data, ViT-B, 16x4) | 81.5 | 95.1 | – | Paper | Code | 2022 | 9 of 13 ran · 4 unverified | report | ||||
| 83 | MoViNet-A6 | 81.5 | 386x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | ||||
| 84 | MLP-3D | 81.4 | – | Paper | – | 2022 | no code linked | report | |||||
| 85 | R[2+1]D-152 (IG-65M pretraining) | 81.3 | 95.1 | – | Paper | Code | 2019 | 1 of 4 ran · 3 unverified | report | ||||
| 86 | LGD-3D Two-stream (ResNet-101) | 81.2 | 95.2 | – | Paper | – | 2019 | no code linked | report | ||||
| 87 | MViT-B, 64x3 | 81.2 | 95.1 | – | Paper | Code | 2021 | 13 of 26 ran · 13 unverified | report | ||||
| 88 | Motionformer-HR | 81.1 | 95.2 | – | Paper | Code | 2021 | 1 of 2 ran · 1 unverified | report | ||||
| 89 | MVD (K400 pretrain, ViT-S, 16x224x224) | 81.0 | 94.8 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 90 | MAR (50% mask, ViT-B, 16x4) | 81.0 | 94.4 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 91 | MoViNet-A5 | 80.9 | 94.9 | 281x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | |||
| 92 | MBT (AV) | 80.8 | 94.6 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 93 | TimeSformer-L | 80.7 | 94.7 | 7140x3 | 121.4 | – | Paper | Code | 2021 | 35 of 43 ran · 8 unverified | report | ||
| 94 | Swin-B (ImageNet-1k pretrain) | 80.6 | 94.6 | – | Paper | Code | 2021 | 7 of 32 ran · 25 unverified | report | ||||
| 95 | Swin-S (ImageNet-1k pretrain) | 80.6 | 94.5 | – | Paper | Code | 2021 | 7 of 32 ran · 25 unverified | report | ||||
| 96 | En-VidTr-L | 80.5 | 94.6 | – | Paper | – | 2021 | no code linked | report | ||||
| 97 | MoViNet-A4 | 80.5 | 94.5 | 105x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | |||
| 98 | OmniSource SlowOnly R101 8x8(ImageNet pretrain) | 80.5 | 94.4 | – | Paper | Code | 2020 | 0 of 13 ran · 13 unverified | report | ||||
| 99 | STAM (64 Frames) | 80.5 | 1040x1 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | ||||
| 100 | X3D-XXL | 80.4 | 94.6 | – | Paper | Code | 2020 | 2 of 15 ran · 13 unverified | report | ||||
| 101 | R3D-RS-200 | 80.4 | 94.4 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 102 | OmniSource SlowOnly R101 8x8 (Scratch) | 80.4 | 94.4 | – | Paper | Code | 2020 | 0 of 13 ran · 13 unverified | report | ||||
| 103 | MViT-B, 32x3 | 80.2 | 94.4 | – | Paper | Code | 2021 | 13 of 26 ran · 13 unverified | report | ||||
| 104 | AMD(ViT-S/16) | 80.1 | 94.5 | 57X15 | 22 | – | Paper | – | 2023 | no code linked | report | ||
| 105 | SlowFast 16x8 (ResNet-101 + NL) | 79.8 | ✓ | Paper | Code | 2018 | 0 of 10 ran · 10 unverified | report | |||||
| 106 | CT-Net Ensemble | 79.8 | – | Paper | Code | 2021 | 10 of 13 ran · 3 unverified | report | |||||
| 107 | ViT-B-VTN+ ImageNet-21K (84.0 [10]) | 79.8 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 108 | TimeSformer-HR | 79.7 | 94.4 | – | Paper | Code | 2021 | 35 of 43 ran · 8 unverified | report | ||||
| 109 | En-VidTr-M | 79.7 | 94.2 | – | Paper | – | 2021 | no code linked | report | ||||
| 110 | LGD-3D RGB (ResNet-101) | 79.4 | 94.4 | – | Paper | – | 2019 | no code linked | report | ||||
| 111 | TDN-ResNet101 (ensemble, ImageNet pretrained, RGB only) | 79.4 | 94.4 | – | Paper | Code | 2020 | 2 of 4 ran · 2 unverified | report | ||||
| 112 | En-VidTr-S | 79.4 | 94 | – | Paper | – | 2021 | no code linked | report | ||||
| 113 | MAR (75% mask, ViT-B, 16x4) | 79.4 | 93.7 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 114 | STAM (16 Frames) | 79.3 | 270x1 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | ||||
| 115 | ip-CSN-152 (Sports-1M pretraining) | 79.2 | 93.8 | – | Paper | Code | 2019 | 1 of 4 ran · 3 unverified | report | ||||
| 116 | CorrNet | 79.2 | – | Paper | – | 2019 | no code linked | report | |||||
| 117 | OmniVL | 79.1 | 94.5 | – | Paper | – | 2022 | no code linked | report | ||||
| 118 | X3D-XL | 79.1 | 93.9 | – | Paper | Code | 2020 | 2 of 15 ran · 13 unverified | report | ||||
| 119 | MVFNet-ResNet101 (ensemble, ImageNet pretrained, RGB only) | 79.1 | 93.8 | – | Paper | Code | 2020 | linked, not harvested | report | ||||
| 120 | TAdaConvNeXt-T | 79.1 | 93.7 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | ||||
| 121 | SlowFast 16x8 (ResNet-101) | 78.9 | 93.5 | – | Paper | Code | 2018 | 0 of 10 ran · 10 unverified | report | ||||
| 122 | G-Blend (Sports-1M pretrain) | 78.9 | – | Paper | Code | 2019 | 1 of 11 ran · 10 unverified | report | |||||
| 123 | Swin-T (ImageNet-1k pretrain) | 78.8 | 93.6 | – | Paper | Code | 2021 | 7 of 32 ran · 25 unverified | report | ||||
| 124 | GB + DF + LB (ResNet 152, ImageNet pretrained) | 78.8 | – | Paper | – | 2019 | no code linked | report | |||||
| 125 | ViT-B-VTN (3 layers, ImageNet pretrain) | 78.6 | 93.7 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 126 | MViT-B, 16x4 | 78.4 | 93.5 | – | Paper | Code | 2021 | 13 of 26 ran · 13 unverified | report | ||||
| 127 | MoViNet-A3 | 78.2 | 93.8 | 56.9x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | |||
| 128 | TAda2D-En (ResNet-50, 8+16 frames) | 78.2 | 93.5 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | ||||
| 129 | SVT | 78.1 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | |||||
| 130 | TimeSformer | 78 | 93.7 | – | Paper | Code | 2021 | 35 of 43 ran · 8 unverified | report | ||||
| 131 | SlowFast 8x8 (ResNet-101) | 77.9 | 93.2 | – | Paper | Code | 2018 | 0 of 10 ran · 10 unverified | report | ||||
| 132 | RepFlow-50 ([2+1]D CNN, FcF, Non-local block) | 77.9 | – | Paper | Code | 2018 | linked, not harvested | report | |||||
| 133 | ip-CSN-152 | 77.8 | 92.8 | – | Paper | Code | 2019 | 1 of 4 ran · 3 unverified | report | ||||
| 134 | I3D + NL | 77.7 | 93.3 | – | Paper | Code | 2017 | 3 of 4 ran · 1 unverified | report | ||||
| 135 | G-Blend | 77.7 | – | Paper | Code | 2019 | 1 of 11 ran · 10 unverified | report | |||||
| 136 | HATNet (32 frames) | 77.6 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 137 | X3D-L | 77.5 | 92.9 | – | Paper | Code | 2020 | 2 of 15 ran · 13 unverified | report | ||||
| 138 | CoST ResNet-101 (ImageNet pretrain) | 77.5 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 139 | TAda2D (ResNet-50, 16 frames) | 77.4 | 93.1 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | ||||
| 140 | EvaNet | 77.4 | – | Paper | – | 2018 | no code linked | report | |||||
| 141 | RNL+TSM Ensemble(ResNet50, 8 + 16 frames) | 77.4 | – | Paper | Code | 2020 | linked, not harvested | report | |||||
| 142 | VIMPAC | 77.4 | – | Paper | Code | 2021 | 1 of 1 ran · 0 unverified | report | |||||
| 143 | BQN (ResNet-50) | 77.3 | 93.2 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 144 | S3D-G (RGB+Flow, ImageNet pretrained) | 77.2 | 93 | – | Paper | Code | 2017 | linked, not harvested | report | ||||
| 145 | SlowFast 8x8 (ResNet-50) | 77 | 92.6 | – | Paper | Code | 2018 | 0 of 10 ran · 10 unverified | report | ||||
| 146 | TAda2D (ResNet-50, 8 frames) | 76.7 | 92.6 | – | Paper | Code | 2021 | 0 of 1 ran · 1 unverified | report | ||||
| 147 | D3D+S3D-G (RGB + RGB) | 76.5 | – | Paper | Code | 2018 | linked, not harvested | report | |||||
| 148 | MSNet-R50 (16 frames, ImageNet pretrained) | 76.4 | – | Paper | Code | 2020 | 3 of 8 ran · 5 unverified | report | |||||
| 149 | GloRe | 76.1 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 150 | X3D-M | 76 | 92.3 | – | Paper | Code | 2020 | 2 of 15 ran · 13 unverified | report | ||||
| 151 | MViT-S | 76 | 92.1 | – | Paper | Code | 2021 | 13 of 26 ran · 13 unverified | report | ||||
| 152 | CMA iter1 (16 frames) | 75.98 | – | Paper | – | 2019 | no code linked | report | |||||
| 153 | D3D (RGB) | 75.9 | – | Paper | Code | 2018 | linked, not harvested | report | |||||
| 154 | Oct-I3D + NL | 75.7 | – | Paper | Code | 2019 | 14 of 34 ran · 20 unverified | report | |||||
| 155 | SlowFast 4x16 (ResNet-50) | 75.6 | 92.1 | – | Paper | Code | 2018 | 0 of 10 ran · 10 unverified | report | ||||
| 156 | R[2+1]D-Flow (Sports-1M pretrain) | 75.4 | 91.9 | – | Paper | Code | 2017 | 1 of 4 ran · 3 unverified | report | ||||
| 157 | FASTER32 | 75.1 | – | Paper | – | 2019 | no code linked | report | |||||
| 158 | MoViNet-A2 | 75.0 | 92.3 | 10.3x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | |||
| 159 | MARS+RGB+Flow (64 frames) | 74.9 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 160 | S3D-G (RGB, ImageNet pretrained) | 74.7 | 93.4 | – | Paper | Code | 2017 | linked, not harvested | report | ||||
| 161 | TSM | 74.7 | – | Paper | Code | 2018 | 6 of 16 ran · 10 unverified | report | |||||
| 162 | A2 Net | 74.6 | 91.5 | – | Paper | – | 2018 | no code linked | report | ||||
| 163 | R[2+1]D-RGB (Sports-1M pretrain) | 74.3 | 91.4 | – | Paper | Code | 2017 | 1 of 4 ran · 3 unverified | report | ||||
| 164 | TSN | 73.9 | 91.1 | – | Paper | Code | 2016 | 2 of 24 ran · 22 unverified | report | ||||
| 165 | R[2+1]D-Two-Stream | 73.9 | 90.9 | – | Paper | Code | 2017 | 1 of 4 ran · 3 unverified | report | ||||
| 166 | TSN | 73.9 | – | Paper | Code | 2017 | linked, not harvested | report | |||||
| 167 | STM (ResNet-50) | 73.7 | – | Paper | – | 2019 | no code linked | report | |||||
| 168 | bLVNet Fan et al. (2019) | 73.5 | 91.2 | – | Paper | Code | 2019 | 0 of 6 ran · 6 unverified | report | ||||
| 169 | Co Slow_64 | 73.05 | 6.90x1 | 32.45 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 170 | Inception-ResNet | 73.0 | 90.9 | – | Paper | – | 2017 | no code linked | report | ||||
| 171 | MFNet | 72.8 | 90.4 | – | Paper | – | 2018 | no code linked | report | ||||
| 172 | MoViNet-A1 | 72.7 | 91.2 | 6.0x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | |||
| 173 | ARTNet | 72.4 | 90.4 | – | Paper | Code | 2017 | linked, not harvested | report | ||||
| 174 | LGD-3D Flow (ResNet-101) | 72.3 | 90.9 | – | Paper | – | 2019 | no code linked | report | ||||
| 175 | R[2+1]D | 72 | 90 | – | Paper | Code | 2017 | 1 of 4 ran · 3 unverified | report | ||||
| 176 | R[2+1]D-RGB | 72 | 90 | – | Paper | Code | 2017 | 1 of 4 ran · 3 unverified | report | ||||
| 177 | FASTER16 w/o sp | 71.7 | – | Paper | – | 2019 | no code linked | report | |||||
| 178 | Co X3D-L_64 | 71.61 | 1.25x1 | 6.15 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 179 | I3D | 71.1 | 89.3 | – | Paper | Code | 2017 | 16 of 28 ran · 12 unverified | report | ||||
| 180 | Co X3D-M_64 | 71.03 | 0.33x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 181 | X3D-L | 69.29 | 19.17x1 | 6.15 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 182 | MARS+RGB+Flow (16 frames) | 68.9 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 183 | SlowFast-8×8-R50 | 68.45 | 66.25x1 | 66.25 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 184 | S3D-G (Flow, ImageNet pretrained) | 68 | 87.6 | – | Paper | Code | 2017 | linked, not harvested | report | ||||
| 185 | R[2+1]D-Flow | 67.5 | 87.2 | – | Paper | Code | 2017 | 1 of 4 ran · 3 unverified | report | ||||
| 186 | Slow-8x8-R50 | 67.42 | 54.87x1 | 32.45 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 187 | Co X3D-S_64 | 67.33 | 0.17x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 188 | X3D-M | 67.24 | 4.97x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 189 | SlowFast-4×16-R50 | 67.06 | 36.46x1 | 34.48 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 190 | Co Slow_8 | 65.90 | 6.90x1 | 32.45 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 191 | MoViNet-A0 | 65.8 | 87.4 | 2.7x1 | – | Paper | Code | 2021 | 8 of 13 ran · 5 unverified | report | |||
| 192 | X3D-S | 64.71 | 2.06x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 193 | I3D-R50 | 63.98 | 28.61x1 | 28.04 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 194 | Co X3D-L_16 | 63.03 | 1.25x1 | 6.15 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 195 | Co X3D-M_16 | 62.80 | 0.33x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 196 | Co X3D-S_13 | 60.18 | 0.17x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 197 | Co I3D_8 | 59.58 | 5.68x1 | 28.04 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 198 | R(2+1)D-18_16 | 59.52 | 40.71x1 | 31.51 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 199 | X3D-XS | 59.37 | 0.64x1 | 3.79 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 200 | Co I3D_64 | 56.86 | 5.68x1 | 28.04 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 201 | R(2+1)D-18_8 | 53.52 | 20.35x1 | 31.51 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 202 | RCU_8 | 53.40 | 4.71x1 | 12.80 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 203 | ViViT-L/16x2 320 | 94.7 | – | Paper | Code | 2021 | 14 of 21 ran · 7 unverified | report | |||||
| 204 | ViT-B-VTN+ ImageNet-21K (84.0 [10]) | 94.2 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 205 | SlowFast 16x8 (ResNet-101 + NL) | 93.9 | – | Paper | Code | 2018 | 0 of 10 ran · 10 unverified | report | |||||
| 206 | ViT-B-VTN (1 layer, ImageNet pretrain) | 93.4 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 207 | MViT-B (train from scratch) | 225x5 | – | Paper | Code | 2021 | linked, not harvested | report |
All 207 rows shown. 207 link to a paper page on this site; 10 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 109 rows have a graph line, from 54 distinct papers; 90 rows (46 papers) have at least one sample that ran. Counting each paper once: Syntology ran 256 of 614 samples; 358 unverified. Separately, 116 of those 614 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections