Tasks SotA Datasets Papers Methods Submit About

Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable Benchmarks All SotA Datasets Papers Methods

Community

Submit Results About

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Models/I3D

I3D

Reported on 27 benchmarks across 12 tasks · 6 papers · 19 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision18 results

Sign Language RecognitiononWLASL100
Top-1 Accuracy· 2019-10-24
65.89
best: 92.25 (Uni-Sign)
SOTA
Word-level Deep Sign Language Recognition from Video: A New Large-scale Dataset and Methods Comparison arXiv:1910.11006
Sign Language RecognitiononWLASL-2000
Top-1 Accuracy· 2019-10-24
32.48
best: 66.82 (Logos-Pretraining)
SOTA
Word-level Deep Sign Language Recognition from Video: A New Large-scale Dataset and Methods Comparison arXiv:1910.11006
VideoonJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
VideoonCharades
MAP· 2017-05-22
32.9
best: 66.3 (TokenLearner)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
VideoonToyota Smarthome dataset
CS· uses extra data· 2017-05-22
53.4
best: 72.9 (π-ViT)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
VideoonToyota Smarthome dataset
CV1· uses extra data· 2017-05-22
34.9
best: 55.2 (π-ViT)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
VideoonToyota Smarthome dataset
CV2· uses extra data· 2017-05-22
45.1
best: 67.8 (EPAM-Net)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
VideoonKinetics-400
Acc@1· 2017-05-22
71.1
best: 93.6 (OmniVec2)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Temporal Action LocalizationonJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Action LocalizationonJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
HandonEgoGesture
Accuracy· 2017-05-22
92.78
best: 94.03 (ResNeXt-101)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
HandonVIVA Hand Gestures Dataset
Accuracy· 2017-05-22
83.1
best: 86.08 (MTUT)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Gesture RecognitiononEgoGesture
Accuracy· 2017-05-22
92.78
best: 94.03 (ResNeXt-101)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Gesture RecognitiononVIVA Hand Gestures Dataset
Accuracy· 2017-05-22
83.1
best: 86.08 (MTUT)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
VideoonMoB
Accuracy· 2023-05-24
72.11
best: 77.85 (VTN)
Malicious or Benign? Towards Effective Content Moderation for Children's Videos arXiv:2305.15551
Video ClassificationonMoB
Accuracy· 2023-05-24
72.11
best: 77.85 (VTN)
Malicious or Benign? Towards Effective Content Moderation for Children's Videos arXiv:2305.15551
VideoonAViD
Accuracy· 2020-07-10
46.8
best: 53.8 (TokenLearner)
AViD Dataset: Anonymized Videos from Diverse Countries arXiv:2007.05515
VideoonKinetics-400
Acc@5· 2017-05-22
89.3
best: 98.9 (TubeViT-H (ImageNet-1k))
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750

Robots4 results

Activity RecognitiononJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Activity RecognitiononHAA500
Top-1 (%)· 2020-09-11
49.87
best: 64.4 (TSN)
HAA500: Human-Centric Atomic Action Dataset with Curated Videos arXiv:2009.05224
Activity RecognitiononMoments in Time Dataset
Top-1 (%)· 2018-01-09
29.51
best: 34.27 (AssembleNet)
Moments in Time Dataset: one million videos for event understanding arXiv:1801.03150
Activity RecognitiononMoments in Time Dataset
Top-5 (%)· 2018-01-09
56.06
best: 62.71 (AssembleNet)
Moments in Time Dataset: one million videos for event understanding arXiv:1801.03150

Time Series3 results

Action DetectiononJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Action RecognitiononJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750
Action RecognitiononHAA500
Top-1 (%)· 2020-09-11
49.87
best: 64.4 (TSN)
HAA500: Human-Centric Atomic Action Dataset with Curated Videos arXiv:2009.05224

Methodology1 result

Zero-Shot LearningonJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750

Natural Language Processing1 result

3D Action RecognitiononJ-HMDB
Accuracy (RGB+pose)· 2017-05-22
84.1
best: 90.4 (Potion)
SOTA
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset arXiv:1705.07750