◉ pulse/lab

Camera frames and results stay in this tab. Save or export a report to keep a copy.

S16-JOVEY-20261026

* Flash status is reported by the browser, not verified by a light sensor.

Automatically selects a usable camera region and colour, then checks distinct beats and waveform stability. Continuously saves clean snippets and reports BP automatically when its signal and model checks pass. SpO₂ is under development and does not delay the report. Brief interruptions discard only the unfinished candidate. Saved counts remain visible; Show available report can finish at any time once a result is available.

Current research settings: at least 12 accepted intervals, two nonoverlapping BP segments, and agreement between earlier and later pulse shapes, intervals, amplitudes and widths. These thresholds are provisional, not a validated minimum recording duration.

BP uses original 2.1-second windows inside nonoverlapping 6.3-second segments. The normalized median waveform is display only. RGB is extracted simultaneously from the whole frame and nine fixed regions. One common region is selected for all three colours, then held fixed while snippets accumulate. RGB waveform collection uses six-second synchronized segments after 4.2 seconds settling for development. Failed windows are retried one second later; accepted snippets never overlap. Every colour must have repeatable beats, matching pulse rates and stable morphology. Original RGB amplitudes and DC levels are retained for development; display normalization does not replace them. Oxygen saturation inference is disabled in this version. More than 5% dark or bright pixels combined in any colour in any frame rejects that RGB segment. Contact, region and sustained exposure changes start a separate group without erasing saved counts. Only compatible snippets within one group are averaged. BP inference uses snippets from the last 60 seconds; RGB development snippets are retained for 120 seconds; older data expires while session saved counts remain. A single brightness spike does not reset the acquisition. Keep the lens and flash covered; if a colour remains clipped, another rear lens may be needed in Details. These are provisional signal checks, not proof of measurement accuracy.

Trace: baseline removed and auto-scaled; raw RGB is preserved. Waveform appearance does not establish BP accuracy.

Oxygen saturation is under development. No SpO₂ estimate is produced.

SpO₂ development research · BP training results

Loading model…

Training results