iBVP Dataset: RGB-Thermal rPPG Dataset with High Resolution Signal Quality Labels

Remote photoplethysmography (rPPG) promises contactless vital sign monitoring across healthcare, human-computer interaction, and wellness applications. Yet a fundamental challenge persists: existing datasets largely neglect ground-truth signal quality assessment, leaving researchers unable to develop algorithms that can reliably distinguish high-quality physiological signals from noise-corrupted data. Our work addresses this gap.

Figure 1: Samples of ground-truth PPG signals from existing datasets showing artifacts in existing dataset.

Dataset Innovation and Technical Specifications

The iBVP dataset directly addresses this critical gap by providing high-resolution signal quality labels alongside comprehensive multi-modal recordings captured across realistic psycho-physiological variations.

Comprehensive Data Acquisition Protocol

Figure 2: Experimental setup showing synchronized RGB-thermal acquisition at 1m distance with ear-based PPG ground truth collection across four distinct psycho-physiological conditions.

The data acquisition protocol induces realistic variations through four carefully designed conditions:

• Condition A: Rhythmic slow breathing and rest (baseline physiological state)
• Condition B: Easy math task (mild cognitive load)
• Condition C: Difficult math task (high cognitive stress)
• Condition D: Guided head movement (motion artifact challenges)

This protocol ensures comprehensive coverage of real-world variations in both physiological states and motion artifacts that rPPG systems must handle in practical deployments.

Signal Quality Labeling

The dataset provides dual signal quality assessment:

Manual Quality Labels (SQ1, SQ2):

• Expert-annotated quality scores (labelled for every 0.5 second segment, extrapolated as per-sample label)
• Ground truth for quality assessment algorithms

SQA-PhysMD Automated Labels:

• Dense per-sample quality scores
• Trained and validated quality assessment model

This dual approach enables both algorithm development and validation for signal quality assessment—a critical capability missing from existing datasets.

Table 1: iBVP Dataset Technical Specifications

Dataset Structure and Organization

Folder Structure

iBVP_Dataset/
├── p01_a/
│   ├── p01_a_rgb/     # RGB frames (.bmp)
│   ├── p01_a_t/       # Thermal frames (.raw)
│   └── p01_a_bvp.csv  # PPG + quality labels
├── p01_b/
│   ├── p01_b_rgb/
│   ├── p01_b_t/
│   └── p01_b_bvp.csv
└── ...

Data Files Content

Table 3: CSV File Structure (pii_x_bvp.csv)

Research Impact and Global Adoption

Since release in April 2024 (until June 2025):

• 40+ research teams globally utilizing the dataset
• 5,000+ article views indicating strong community engagement
• Multiple high-impact publications leveraging signal quality annotations

Dataset Access and Usage Guidelines

Academic Access Requirements

The iBVP dataset is available exclusively for academic research purposes through a formal request process:

1. Download EULA: Complete the End-User License Agreement
2. Academic Supervisor Signature: In case of students, EULA must be signed by academic supervisors
3. Research Description: Include research objectives and dataset usage plans

Privacy and Ethical Considerations

Confidential Participants: Seven participants (p08, p10, p13, p16, p29, p31, p33) provided limited consent:

• Their data is stored in separate “Confidential_No-media-use” folder
• Images of these participants SHALL NOT be used for any publication.

Conclusion

With comprehensive multi-modal recordings, realistic experimental protocols, and signal quality labeling, iBVP dataset enables the development of robust contactless monitoring systems ready for real-world deployment.

Dataset Access: Submit signed EULA to request academic access

Paper: Electronics 2024

Citation:

@article{joshi2024ibvp,
    title={iBVP Dataset: RGB-Thermal rPPG Dataset with High Resolution Signal Quality Labels},
    author={Joshi, Jitesh and Cho, Youngjun},
    journal={Electronics},
    publisher={MDPI},
    volume={13},
    year={2024},
    number={7},
    article-number={1334},
    url={https://www.mdpi.com/2079-9292/13/7/1334},
    issn={2079-9292},
    doi={10.3390/electronics13071334}
}

@article{joshi2023physiokit,
    title={PhysioKit: An Open-Source, Low-Cost Physiological Computing Toolkit for Single-and Multi-User Studies},
    author={Joshi, Jitesh and Wang, Katherine and Cho, Youngjun},
    journal={Sensors},
    publisher={MDPI},
    volume={23},
    number={19},
    article-number={8244},
    year={2023},
    url={https://www.mdpi.com/1424-8220/23/19/8244},
    issn={1424-8220},
    doi={10.3390/s23198244}
}

This work was conducted at the Department of Computer Science, University College London, under the supervision of Prof. Youngjun Cho. Jitesh Joshi was fully supported with international studentship that was secured by Prof. Cho

For questions or collaborations, please contact: jitesh.joshi.20@ucl.ac.uk and/or youngjun.cho@ucl.ac.uk