发表机构
National Cheng Kung University（成功大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对RhythmFormer远程光体积描记术，量化对比多种可解释AI方法，发现Beyond Intuition在跨数据集上表现最优，其皮肤覆盖率与部分性能指标相关，而SaCo与性能无直接关联。
AI 中文摘要
背景：远程光体积描记术可从面部视频中心血管脉搏，其解释依赖于热力图检查而非模型读取位置的定量证据。我们对这些解释进行量化，探究其是否可在数据集间迁移并跟踪模型性能。方法：我们在NCKU-rPPG数据集上训练8种特定条件的RhythmFormer模型，该数据集在3种光照水平、说话、旋转和骑行场景下采集，每5.12秒片段估计1个心率，并与UBFC-rPPG复制品对比。我们通过皮肤覆盖率和显著性引导保真系数（SaCo）评估原始注意力、rollout、注意力流及Beyond Intuition方法。结果：Beyond Intuition在两个数据集上排名最高，静态3级的中位数覆盖率为0.789、SaCo为0.837，对应UBFC-rPPG上为0.826和0.917；其他方法排名较低。在单个条件的单个参与者内，两种指标与片段的心率误差、波形相关性或信噪比均无关联：252个系数中186个低于|ρ|=0.10，仅28个达到p<0.05（仅偶然预期13个）。8种场景中，仅Beyond Intuition的覆盖率与3种性能指标相关，ρ分别为-0.43、+0.57和+0.43，而仅注意力方法的SaCo与各指标相反。该方法仅在40 lux时失效，中位数覆盖率降至0.180、中位数SaCo降至-0.178，而运动对估计的退化远大于该下降。结论：皮肤覆盖率和SaCo携带与性能指标互补的信息，而非其代理：归因于皮肤不保证估计准确。归因揭示的是模型在某条件下的关注位置，而非其映射的保真度。
英文摘要
Background. Remote photoplethysmography estimates the cardiovascular pulse from facial video, and its explanations have rested on inspecting heatmaps rather than on quantitative evidence about where a model reads it. We quantified the explanations and asked whether such explanations transfer between datasets and track model performance. Method. We trained eight condition-specific RhythmFormer models on NCKU-rPPG, recorded under three illumination levels, speaking, rotation, and cycling, estimated one heart rate per 5.12-second clip, and set them beside a UBFC-rPPG reproduction. Raw attention, rollout, attention flow, and Beyond Intuition were assessed by skin coverage and the Salience-guided Faithfulness Coefficient (SaCo). Results. Beyond Intuition ranked highest on both datasets, at median coverage 0.789 and SaCo 0.837 on Static level 3 against 0.826 and 0.917 on UBFC-rPPG; lower ranks differed. Within one participant of one condition, neither measure was related to a clip's heart-rate error, waveform correlation, or signal-to-noise ratio on either dataset: 186 of the 252 coefficients fell below $|ρ|=0.10$ and 28 reached $p<0.05$ against the 13 expected by chance. Across the eight scenarios only Beyond Intuition's coverage followed the three performance measures, at $ρ=-0.43$, $+0.57$, and $+0.43$, while the attention-only methods' SaCo ran opposite to each. It failed at 40 lux alone, its median coverage falling to 0.180 and its median SaCo to $-0.178$, whereas motion degraded the estimates far more without such a drop. Conclusions. Skin coverage and SaCo carry information complementary to the performance measures rather than a proxy for them: attributing to the skin does not guarantee an accurate estimate. What an attribution reveals about a condition is where the model looks rather than how faithfully its map is ordered.
Comments92 pages, 38 figures, incl. supplementary
