Heart Rate Estimation on PURE
0.12MAERhythmJEPA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RhythmJEPAVenue=—2026.06 | 0.12 | 0.28 | 1 | 0.18 | |
| PHASE-NetVenue=CVPR’262026.04 | 0.14 | 0.35 | 0.99 | — | |
| LSTSVenue=TCSVT'252026.06 | 0.15 | 0.4 | 0.99 | 0.22 | |
| PhysLLMMode=Intra-dataset testing2025.05 | 0.17 | 0.35 | 0.99 | — | |
| PhysLLMVenue=ICLR’262026.04 | 0.17 | 0.35 | 0.99 | — | |
| FreqPhysVenue=-2026.04 | 0.17 | 0.25 | 1 | — | |
| LQ-rPPG2026.05 | 0.18 | 0.28 | 0.99 | — | |
| RhythmMambaInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 0.23 | 0.34 | 0.99 | — | |
| RhythmMambaVenue=AAAI’252026.04 | 0.23 | 0.34 | 0.99 | — | |
| RhythmMamba2026.05 | 0.23 | 0.34 | 0.99 | — | |
| FacePhysTraining dataset=RLAP2025.12 | 0.24 | 0.68 | 1 | — | |
| RhythmFormerInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 0.27 | 0.47 | 0.99 | — | |
| MeshPhysInput Representation=STGRAPH, Supervision Type=Supervised2026.01 | 0.27 | 0.48 | 1 | — | |
| RhythmFormerMode=Intra-dataset testing2025.05 | 0.27 | 0.47 | 0.99 | — | |
| RhythmFormer2026.05 | 0.27 | 0.47 | 0.99 | — | |
| RhythmFormerVenue=PR'252026.06 | 0.27 | 0.46 | 1 | 0.31 | |
| PhysDiffVenue=AAAI’252026.04 | 0.29 | 0.54 | 1 | — | |
| RhythmMambaVenue=AAAI'252026.06 | 0.29 | 0.4 | 0.99 | 0.36 | |
| TS-CANVenue=NeurIPS'202026.06 | 0.32 | 0.63 | 0.99 | 0.5 | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=TAS, Quality Criterion=rPPG-VQA2026.04 | 0.36 | 0.44 | 1 | — | |
| POSVenue=TBME'172026.06 | 0.36 | 0.93 | 1 | 0.5 | |
| Dual-TLInput Representation=STMAPS, Supervision Type=Supervised2026.01 | 0.37 | 0.68 | 0.99 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=rPPG-VQA2026.04 | 0.37 | 0.47 | 1 | — | |
| RIS-iPPGType=Data-Driver, Supervised2026.04 | 0.38 | 1.28 | 0.99 | — | |
| DRNet2022.06 | 0.39 | 0.52 | 0.998 | — | |
| CodePhysInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 0.39 | 0.83 | 0.99 | — | |
| CodePhysVenue=JBHI’252026.04 | 0.39 | 0.83 | 0.99 | — | |
| CodePhysSource=Excerpted from original paper2026.05 | 0.39 | 0.83 | 0.99 | — | |
| Style-rPPGSource=Excerpted from original paper2026.05 | 0.39 | 0.62 | 0.99 | — | |
| rPPG-MAEInput Representation=STMAPS, Supervision Type=Self-supervised2026.01 | 0.4 | 0.92 | 0.99 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=rPPG-VQA2026.04 | 0.42 | 0.5 | 1 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=SNR2026.04 | 0.42 | 0.55 | 1 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=COVER2026.04 | 0.46 | 0.6 | 1 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=SNR2026.04 | 0.47 | 0.58 | 1 | — | |
| Contrast-PhysTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=TAS, Quality Criterion=rPPG-VQA2026.04 | 0.47 | 0.74 | 1 | — | |
| Contrast-Phys+Mode=Intra-dataset testing2025.05 | 0.48 | 0.98 | 0.99 | — | |
| Contrast-Phys+Venue=TPAMI’242026.04 | 0.48 | 0.98 | 0.99 | — | |
| ContrastPhys-100Type=Model-Based Unsupervised2026.04 | 0.48 | 0.98 | 0.99 | — | |
| Contrast-Phys+2026.05 | 0.48 | 0.98 | 0.99 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=COVER2026.04 | 0.5 | 0.58 | 1 | — | |
| PhysFormerVenue=CVPR'222026.06 | 0.52 | 1.03 | 0.99 | 0.86 | |
| EfficientPhysVenue=WACV'232026.06 | 0.55 | 1.34 | 0.99 | 0.71 | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=IPR2026.04 | 0.57 | 0.75 | 1 | — | |
| Gideon21Training Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=IPR2026.04 | 0.57 | 0.72 | 1 | — | |
| PhysNetTraining dataset=RLAP2025.12 | 0.63 | 2.34 | 1 | — | |
| Li et al.Venue=ICCV’232026.04 | 0.64 | 1.16 | 0.99 | — | |
| Li et al.2026.05 | 0.64 | 1.16 | 0.99 | — | |
| STGNetInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 0.69 | 0.98 | 0.99 | — | |
| Contrast-PhysTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=rPPG-VQA2026.04 | 0.71 | 1.6 | 0.99 | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=TAS, Quality Criterion=rPPG-VQA2026.04 | 0.72 | 3.43 | 0.97 | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=rPPG-VQA2026.04 | 0.74 | 4.12 | 0.96 | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=SNR2026.04 | 0.76 | 4.23 | 0.95 | — | |
| Dual-GANReference=CVPR2022.06 | 0.82 | 1.31 | 0.99 | — | |
| Dual-GANInput Representation=STMAPS, Supervision Type=Supervised2026.01 | 0.82 | 1.31 | 0.99 | — | |
| Dual-GANVenue=CVPR’212026.04 | 0.82 | 1.31 | 0.99 | — | |
| DualGANType=Model-Based Unsupervised2026.04 | 0.82 | 1.31 | 0.99 | — | |
| Dual-GAN2026.05 | 0.82 | 1.31 | 0.99 | — | |
| DeepPhysInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 0.83 | 1.54 | 0.99 | — | |
| DeepPhysVenue=ECCV’182026.04 | 0.83 | 1.54 | 0.99 | — | |
| DeepPhys2026.05 | 0.83 | 1.54 | 0.99 | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=rPPG-VQA2026.04 | 0.87 | 4.33 | 0.95 | — | |
| Contrast-PhysTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=SNR2026.04 | 0.88 | 1.76 | 0.99 | — | |
| PhysFormerTraining dataset=RLAP2025.12 | 0.91 | 3.32 | 0.99 | — | |
| PhysNetVenue=BMVC'192026.06 | 0.93 | 2.08 | 0.99 | 1.4 | |
| Contrast-PhysVenue=ECCV’222026.04 | 1 | 1.4 | 0.99 | — | |
| ContrastPhys-0Type=Data-Driven Unsupervised2026.04 | 1 | 1.4 | 0.99 | — | |
| Contrast-Phys+Venue=TPAMI'242026.06 | 1 | 1.4 | 0.99 | 1.21 | |
| PhysFormerInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 1.1 | 1.75 | 0.99 | — | |
| PhysFormerMode=Intra-dataset testing2025.05 | 1.1 | 1.75 | 0.99 | — | |
| PhysFormerVenue=CVPR’222026.04 | 1.1 | 1.75 | 0.99 | — | |
| PhysFormer2026.05 | 1.1 | 1.75 | 0.99 | — | |
| EfficientPhys-T12021.10 | 1.11 | 5.94 | 0.97 | 1.3 | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=SNR2026.04 | 1.19 | 4.97 | 0.94 | — | |
| Yue et al.Venue=TPAMI’232026.04 | 1.23 | 2.01 | 0.99 | — | |
| YueType=Data-Driven Unsupervised2026.04 | 1.23 | 2.01 | 0.99 | — | |
| CVDInput Representation=STMAPS, Supervision Type=Supervised2026.01 | 1.29 | 2.01 | 0.98 | — | |
| CVDVenue=ECCV’202026.04 | 1.29 | 2.01 | 0.98 | — | |
| CVDType=Model-Based Unsupervised2026.04 | 1.29 | 2.01 | 0.98 | — | |
| EfficientPhys-C2021.10 | 1.33 | 5.99 | 0.97 | 1.71 | |
| EfficientPhys2026.05 | 1.33 | 5.99 | 0.97 | — | |
| Contrast-PhysTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=rPPG-VQA2026.04 | 1.36 | 4.05 | 0.96 | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=COVER2026.04 | 1.39 | 5.46 | 0.92 | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=IPR2026.04 | 1.39 | 6.67 | 0.88 | — | |
| TFA-PFE2026.05 | 1.44 | 2.5 | — | — | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=IPR2026.04 | 1.52 | 6.54 | 0.89 | — | |
| RhyMambaTraining dataset=RLAP2025.12 | 1.63 | 4.55 | 0.98 | — | |
| Contrast-PhysTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=WRS, Quality Criterion=COVER2026.04 | 1.64 | 7.17 | 0.86 | — | |
| POS2021.10 | 1.68 | 9.6 | 0.92 | 1.56 | |
| SiNCTraining Set=CAS(ME)2 + Celeb-DF (v2), Sampling=Top/Bottom K, Quality Criterion=COVER2026.04 | 1.69 | 6.76 | 0.88 | — | |
| Contrast-PhysTraining Set=CAS(ME)2, Sampling=-, Quality Criterion=-2026.04 | 1.72 | 2.74 | 0.98 | — | |
| Gideon21Training Set=CAS(ME)2, Sampling=-, Quality Criterion=-2026.04 | 1.8 | 2.24 | 0.99 | — | |
| HR-CNNReference=BMVC2022.06 | 1.84 | 2.37 | 0.98 | — | |
| HR-CNNType=Model-Based Unsupervised2026.04 | 1.84 | 2.37 | — | — | |
| HR-CNN2026.05 | 1.84 | 2.37 | 0.98 | — | |
| SiNCTraining Set=CAS(ME)2, Sampling=-, Quality Criterion=-2026.04 | 2 | 8.16 | 0.93 | — | |
| CHROMReference=TBME2022.06 | 2.07 | 9.92 | 0.99 | — | |
| CHROMVenue=TBE’132026.04 | 2.07 | 2.5 | 0.99 | — | |
| CHROMType=Model-Based Unsupervised2026.04 | 2.07 | 9.92 | — | — | |
| CHROM2026.05 | 2.07 | 2.5 | 0.99 | — | |
| PhysNetInput Representation=VIDEO, Supervision Type=Supervised2026.01 | 2.1 | 2.6 | 0.99 | — |