SeetaPsych Hertz

Heart rate estimation modules for SeetaPsych

Usage

This project is already included in the seetapsych-lib default configuration. Download and use it via seetapsych-manager download.

For usage, refer to SeetaPsych.

You can additionally add this algorithm module using the following methods.

Heart rate estimation requires processing video or real-time video streams to extract heart rate information.

WebUI

Run seetapsych-webui with the --files argument to use it.

seetapsych-webui --files seetapsych_hertz/modules/seeta.yml

Programmatic Usage

Add the following code in your program to use this algorithm module.

from seetapsych_lib.runtime.factory import Factory
from seetapsych_lib.runtime.pipeline import Pipeline

factory = Factory()
factory.load_file_modules('seetapsych_hertz/modules/seeta.yml')

pipeline = Pipeline(factory, ...)

pipeline.add_attributes('face/heart_rate')

Introduction

SeetaHeartRateDetector (Signal-based)

Traditional signal-processing-based heart rate estimation. Extracts ROI from facial dense landmarks, then applies chrominance analysis with cubic spline interpolation and FFT frequency-domain peak detection to estimate heart rate from BGR channel signals.

Module config: seeta.yml. Provide Attributes: face/heart_rate.

Requires: face/dense_landmarks.

Parameters:

• min_seconds (float, default 1): Minimum signal duration in seconds before estimation starts.
• min_frames (int, default 10): Minimum number of frames before estimation starts.
• max_frames (int, default 300): Maximum frames kept in the sliding signal window.

AdaChrom (Adaptive Chrominance)

Adaptive chrominance-based heart rate estimation using frame-wise sliding window analysis. Extracts photoplethysmographic signals from an adaptive forehead skin ROI, processes each frame with chrominance methods to produce real-time BPM output without FFT fusion.

Module config: ada-chrom.yml. Provide Attributes: face/heart_rate.

Requires: face/dense_landmarks.

Parameters:

• window_samples (int, default 300): Number of frames in the sliding estimation window.

TinyHR (ONNX-based)

RhythmFormer ONNX heart rate estimation engine. Converts face-frame sequences into BVP signals using a deep neural network, then applies long-window BVP fusion and Welch spectral analysis for stable heart-rate output.

Module config: tiny-hr.yml. Provide Attributes: face/heart_rate.

Requires: face/detection.

Available model: seeta-hertz-tinyhr.onnx.

Parameters:

• fps (float, default 30): Expected camera/video FPS.
• interval (float, default 1): Interval in seconds between heart rate estimations.