An older case I prepared for an interview, most of the work follows the case’s specifications, but feel free to grab anything you might find useful.
Link to the original paper and the dataset: Paper Dataset
The task is to implement a heart rate estimation model based on the Deep PPG paper, using the PPG-DaLiA dataset, which includes:
- ECG data (ground truth for heart rate)
- PPG + accelerometer data (for motion-compensated HR estimation)
The model follows a CNN-based approach as outlined in Section 4 of the paper.
This repo contains:
- Data preprocessing & visualization (LOSO cross-validation setup)
- CNN model implementation & training
- Error analysis & evaluation
Clone the repo and install dependencies.
git clone https://github.com/deep_ppg_hr.git
cd deep_ppg_hr
pip install -r requirements.txt