Pulse Rate Estimation on VIPL-HR

3.53MAE (BPM)MeshPhys

Evaluation results

Method
MeshPhys2026.01	3.53	6.82	0.84
FreqPhys2026.04	3.79	6.34	0.86
PhysDiff2026.04	3.92	6.65	0.85
RhythmGaussian2026.01	4.22	7.12	0.92
PhysLLM2026.04	4.24	6.81	—
CodePhys2026.01	4.27	7.11	0.81
CodePhys2026.04	4.27	7.11	0.81
RhythmMamba2026.01	4.3	7.49	0.81
RhythmMamba2026.04	4.3	7.49	0.81
Proposed (WMST + Wavelet-aware contrastive learning)2026.03	4.32	7.26	0.86
Dual-TL2026.01	4.36	6.92	0.69
RS+rPPG2026.03	4.4	7.19	0.79
RhythmFormer2026.01	4.51	7.98	0.78
rPPG-MAE2026.01	4.52	7.49	0.81
PhysGCN2026.01	4.7	7.44	0.82
NEST2026.01	4.76	7.51	0.84
PhysFormer++2026.01	4.88	7.62	0.8
Dual-GAN2026.01	4.93	7.68	0.81
Dual-GAN2026.04	4.93	7.68	0.81
PhysFormer2026.01	4.97	7.79	0.78
PhysFormer2026.04	4.97	7.79	0.78
Li et al.2026.04	4.97	7.79	0.78
CVD2026.01	5.02	7.97	0.79
CVD2026.04	5.02	7.97	0.79
NAS-HR2026.01	5.12	8.01	0.79
Li et al.2026.03	5.19	8.26	0.78
RhythmNet2026.01	5.3	8.14	0.76
BVPNet2026.01	5.34	7.85	0.7
RS-rPPG.2026.03	5.98	10.5	0.56
SSPD2026.03	6.04	9.56	0.59
GraphPhys2026.01	6.69	9.7	0.48
Greip et al.2026.03	7.35	9.7	0.55
Gideon et al.2026.04	9.01	14.02	0.58
PhysNet2026.01	10.8	14.8	0.2
PhysNet2026.04	10.8	14.8	0.2
DeepPhys2026.01	11	13.8	0.11
DeepPhys2026.04	11	13.8	0.72
CHROM2026.04	11.4	16.9	0.28
CHROM2026.01	11.44	16.97	0.28
POS2026.01	11.5	17.2	0.3
Contrast-Phys2026.03	17.8	22.8	0.17
Contrast-Phys2026.04	32.1	36.1	0.04