Liu et al.

Reported on 30 benchmarks across 13 tasks · 5 papers · 16 SOTA

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

Computer Vision23 results

inverse tone mappingonVDS dataset: Multi exposure stack-based inverse tone mapping
HDR-VDP-2· 2020-04-02
56.97
best: 59 (CEVR)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
inverse tone mappingonVDS dataset: Multi exposure stack-based inverse tone mapping
HDR-VDP-3· 2020-04-02
8.24
best: 8.95 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
inverse tone mappingonVDS dataset: Multi exposure stack-based inverse tone mapping
Kim and Kautz TMO-PSNR· 2020-04-02
28
best: 30.04 (CEVR)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
inverse tone mappingonVDS dataset: Multi exposure stack-based inverse tone mapping
PU21-PSNR· 2020-04-02
25.69
best: 31.15 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
inverse tone mappingonVDS dataset: Multi exposure stack-based inverse tone mapping
PU21-SSIM· 2020-04-02
0.8797
best: 0.9537 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
inverse tone mappingonVDS dataset: Multi exposure stack-based inverse tone mapping
Reinhard'TMO-PSNR· 2020-04-02
30.88
best: 35.75 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
Inverse-Tone-MappingonVDS dataset: Multi exposure stack-based inverse tone mapping
HDR-VDP-2· 2020-04-02
56.97
best: 59 (CEVR)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
Inverse-Tone-MappingonVDS dataset: Multi exposure stack-based inverse tone mapping
HDR-VDP-3· 2020-04-02
8.24
best: 8.95 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
Inverse-Tone-MappingonVDS dataset: Multi exposure stack-based inverse tone mapping
Kim and Kautz TMO-PSNR· 2020-04-02
28
best: 30.04 (CEVR)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
Inverse-Tone-MappingonVDS dataset: Multi exposure stack-based inverse tone mapping
PU21-PSNR· 2020-04-02
25.69
best: 31.15 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
Inverse-Tone-MappingonVDS dataset: Multi exposure stack-based inverse tone mapping
PU21-SSIM· 2020-04-02
0.8797
best: 0.9537 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
Inverse-Tone-MappingonVDS dataset: Multi exposure stack-based inverse tone mapping
Reinhard'TMO-PSNR· 2020-04-02
30.88
best: 35.75 (Deep Conditional HDRI)
SOTA
Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline arXiv:2004.01179
3D Multi-Person Pose Estimation (root-relative)onMuPoTS-3D
3DPCK· 2022-07-29
79.4
best: 89.6 (TDBU_Net)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
3D Human Pose EstimationonMuPoTS-3D
3DPCK· 2022-07-29
36.5
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
3D Human Pose EstimationonMuPoTS-3D
3DPCK· 2022-07-29
79.4
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
3D Multi-Person Pose Estimation (absolute)onMuPoTS-3D
3DPCK· 2022-07-29
36.5
best: 50.9 (POTR-3D)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
Pose EstimationonMuPoTS-3D
3DPCK· 2022-07-29
36.5
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
Pose EstimationonMuPoTS-3D
3DPCK· 2022-07-29
79.4
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
3D Multi-Person Pose EstimationonMuPoTS-3D
3DPCK· 2022-07-29
36.5
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
3D Multi-Person Pose EstimationonMuPoTS-3D
3DPCK· 2022-07-29
79.4
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
CrowdsonShanghaiTech B
MAE· 2018-03-08
13.7
best: 5.51 (EBC-ZIP-B)
Leveraging Unlabeled Data for Crowd Counting by Learning to Rank arXiv:1803.03095
CrowdsonShanghaiTech A
MAE· 2018-03-08
73.6
best: 47.81 (EBC-ZIP-B)
Leveraging Unlabeled Data for Crowd Counting by Learning to Rank arXiv:1803.03095
CrowdsonUCF CC 50
MAE· 2018-03-08
337.6
best: 154.8 (APGCC)
Leveraging Unlabeled Data for Crowd Counting by Learning to Rank arXiv:1803.03095

Natural Language Processing4 results

Sentiment AnalysisonSentihood
Aspect· 2018-04-30
78.5
best: 87.9 (BERT-pair-QA-B)
SOTA
Recurrent Entity Networks with Delayed Memory Update for Targeted Aspect-based Sentiment Analysis arXiv:1804.11019
Sentiment AnalysisonSentihood
Sentiment· 2018-04-30
91
best: 93.6 (BERT-pair-QA-M)
SOTA
Recurrent Entity Networks with Delayed Memory Update for Targeted Aspect-based Sentiment Analysis arXiv:1804.11019
Aspect-Based Sentiment Analysis (ABSA)onSentihood
Aspect· 2018-04-30
78.5
best: 87.9 (BERT-pair-QA-B)
SOTA
Recurrent Entity Networks with Delayed Memory Update for Targeted Aspect-based Sentiment Analysis arXiv:1804.11019
Aspect-Based Sentiment Analysis (ABSA)onSentihood
Sentiment· 2018-04-30
91
best: 93.6 (BERT-pair-QA-M)
SOTA
Recurrent Entity Networks with Delayed Memory Update for Targeted Aspect-based Sentiment Analysis arXiv:1804.11019

Speech4 results

DialogueonSecond dialogue state tracking challenge
Area· 2018-04-18
90
best: 92 (RNN)
Dialogue Learning with Human Teaching and Feedback in End-to-End Trainable Task-Oriented Dialogue Systems arXiv:1804.06512
DialogueonSecond dialogue state tracking challenge
Food· 2018-04-18
84
best: 86 (RNN)
Dialogue Learning with Human Teaching and Feedback in End-to-End Trainable Task-Oriented Dialogue Systems arXiv:1804.06512
DialogueonSecond dialogue state tracking challenge
Joint· 2018-04-18
72
best: 85 (Seq2Seq-DU-w/oSchema)
Dialogue Learning with Human Teaching and Feedback in End-to-End Trainable Task-Oriented Dialogue Systems arXiv:1804.06512
DialogueonSecond dialogue state tracking challenge
Price· 2018-04-18
92
best: 94 (Neural belief tracker)
Dialogue Learning with Human Teaching and Feedback in End-to-End Trainable Task-Oriented Dialogue Systems arXiv:1804.06512

Methodology2 results

3DonMuPoTS-3D
3DPCK· 2022-07-29
36.5
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
3DonMuPoTS-3D
3DPCK· 2022-07-29
79.4
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090

Audio2 results

1 Image, 2*2 StitchionMuPoTS-3D
3DPCK· 2022-07-29
36.5
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090
1 Image, 2*2 StitchionMuPoTS-3D
3DPCK· 2022-07-29
79.4
best: 89.9 (Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement)
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation arXiv:2208.00090