AI Paper Atlas

Research map / Benchmarks / THUMOS14

THUMOS14 leaderboard

74 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on THUMOS14. Best mAP@0.5: 74.4 (MDVLM-TAL-8B (ours), NeurIPS 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

THUMOS14 · mAP@0.5

23 papers, each with its best reported number for its own method; higher is better.

13 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

THUMOS14 · average mAP

21 papers, each with its best reported number for its own method; higher is better.

11 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

THUMOS14 · mAP@IoU0.5

18 papers, each with its best reported number for its own method; higher is better.

8 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

THUMOS14 · mAP

12 papers, each with its best reported number for its own method; higher is better.

#mAP ↑MethodPaperTopic
173.2CMeRTContext-Enhanced Memory-Refined Transformer for Online Action Detection
CVPR 2025 · Table 2
Action recognition
272.5GateHUB
TimeSformer RGB + TSN optical flow
GateHUB: Gated History Unit With Background Suppression for Online Action Detection
CVPR 2022 · Table 1
Action recognition
372.4E2E-LOAD
MViT-style, RGB only
E2E-LOAD: End-to-End Long-form Online Action Detection
ICCV 2023 · Table 1
Action recognition
469.6VideoMAE V2
ViT-g
VideoMAE V2: Scaling Video Masked Autoencoders With Dual Masking
CVPR 2023 · Table 5 (THUMOS14 table)
Video segmentation and tracking
561.5EMV+RGB-CNNReal-Time Action Recognition With Enhanced Motion Vector CNNs
CVPR 2016 · Table: Comparison of speed and
Action recognition
655.2ALBAR features + TriDet
Swin-T finetuned on HMDB51
ALBAR: Adversarial Learning approach to mitigate Biases in Action Recognition
ICLR 2025 · Downstream task table
Action recognition
747.2TRNTemporal Recurrent Networks for Online Action Detection
ICCV 2019 · Table 3
Action recognition
845.1I3D, 10 FPS, feature fusion
I3D + TadTR
An Empirical Study of End-to-End Temporal Action Detection
CVPR 2022 · Appendix Table: Effect of mult
Action recognition
938.2GTAN
P3D
Gaussian Temporal Awareness Networks for Action Localization
CVPR 2019 · Table 3
Action recognition
1037.5OV-OAD (ANet)
CLIP ViT/B
Does Video-Text Pretraining Help Open-Vocabulary Online Action Detection?
NeurIPS 2024 · Table 1
Action recognition
2 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

THUMOS14 · avg mAP

11 papers, each with its best reported number for its own method; higher is better.

1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

THUMOS14 · mAP@IoU0.3

11 papers, each with its best reported number for its own method; higher is better.

#mAP@IoU0.3 ↑MethodPaperTopic
161.5Ours
I3D + CLIP
Distilling Vision-Language Pre-Training To Collaborate With Weakly-Supervised Temporal Action Localization
CVPR 2023 · Table 1
Action recognition
256.5DELU
CO2-Net backbone, I3D features
Dual-Evidential Learning for Weakly-Supervised Temporal Action Localization
ECCV 2022 · Table 1
Action recognition
356.2OursBoosting Weakly-Supervised Temporal Action Localization With Text Information
CVPR 2023 · Table 1
Action recognition
452.3D2-Net
I3D
D2-Net: Weakly-Supervised Action Localization via Discriminative Embeddings and Denoised Activations
ICCV 2021 · Table 1
Action recognition
548.2Ours (I3D)
I3D
Relational Prototypical Network for Weakly Supervised Temporal Action Localization
AAAI 2020 · Table 1
Action recognition
648.1A2CL-PT
I3D features
Adversarial Background-Aware Loss for Weakly-supervised Temporal Activity Localization
ECCV 2020 · Table 1
Action recognition
739.5TSM (I3D, 3-phase)
I3D
Temporal Structure Mining for Weakly Supervised Action Detection
ICCV 2019 · Table 5
Action recognition
838.3TSRNet
ResNet101
Learning Transferable Self-Attentive Representations for Action Recognition in Untrimmed Videos with Weak Supervision
AAAI 2019 · Table: Comparisons on THUMOS14
Action recognition
936Ours (full)End-To-End Learning of Action Detection From Frame Glimpses in Videos
CVPR 2016 · THUMOS'14 results table
Action recognition
1035.8AutoLoc (weak)AutoLoc: Weakly-supervised Temporal Action Localization in Untrimmed Videos
ECCV 2018 · Table 1
Action recognition
1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

THUMOS14 · mAP@0.7

11 papers, each with its best reported number for its own method; higher is better.

1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.