PRPIP is a project focuses on the reconstruction of pupil size during blink intervals in eye-tracking data, leveraging a physiologically inspired approach to enhance data continuity and accuracy. The core objective is to address the challenge posed by data gaps introduced by blinks, which are common in eye-tracking recordings, by applying methods rooted in the natural dynamics of pupil behavior.
The PRPIP method automatically identifies blink intervals within the dataset, providing an efficient preprocessing step that eliminates the need for manual intervention. The reconstruction process utilizes a logarithmic recovery model, dynamically adjusting the recovery time constant () to simulate the natural re-dilation of the pupil post-blink. To reflect the inherent variability observed in physiological measurements, the algorithm integrates Gaussian noise into the reconstruction process, ensuring the generated data maintains biological plausibility.
The PRPIP framework emphasizes data integrity and user accessibility. It validates the input data for compatibility, offering intuitive and detailed error messages to assist users in correcting potential issues. The system supports a wide range of input formats, including CSV, Excel, JSON, Parquet, and TXT, ensuring flexibility in its application. Output options are designed to accommodate different analytical needs, allowing users to either append a new column for reconstructed pupil size or replace the original column directly with reconstructed values.
This approach not only enhances the utility of eye-tracking datasets by mitigating the impact of blink-induced artifacts but also aligns closely with the physiological characteristics of pupil dynamics, ensuring that the reconstructed data remains consistent with real-world observations. This combination of automated detection, physiologically inspired modeling, and user-friendly design makes the project a valuable tool for researchers and practitioners in the field of eye-tracking and visual behavior analysis.
To access the full guide and content of PRPIP, you can explore the following resources:
The permanent GitHub repository provides comprehensive documentation and code. Additionally, PRPIP is available on the PyPI index, making it easy to install via pip install prpip. These platforms ensure seamless access to PRPIP’s functionality and updates.