A Step Forward in Infant Eye Tracking: Capturing Real-World Gaze Data for Young Children
Research Digest
June 17, 2025
Figure showing an infant seated on their parent’s lap, wearing a Pupil Labs Neon eye tracking module mounted on a headband (lower panel). The top panel displays the live visual feedback from the sensor module’s camera, as seen through the Neon Companion App. Source: Using Pupil Labs' Neon eye-tracking module for infants aged 3 to 24 months.
Figure showing an infant seated on their parent’s lap, wearing a Pupil Labs Neon eye tracking module mounted on a headband (lower panel). The top panel displays the live visual feedback from the sensor module’s camera, as seen through the Neon Companion App. Source: Using Pupil Labs' Neon eye-tracking module for infants aged 3 to 24 months.
Figure showing an infant seated on their parent’s lap, wearing a Pupil Labs Neon eye tracking module mounted on a headband (lower panel). The top panel displays the live visual feedback from the sensor module’s camera, as seen through the Neon Companion App. Source: Using Pupil Labs' Neon eye-tracking module for infants aged 3 to 24 months.
The Challenges of Infant Eye Tracking
Studying how infants visually engage with their environment is crucial for understanding early cognitive development. Historically, research has depended on lab-based eye tracking setups or video analysis from head-mounted cameras, which often restrict studies to controlled settings and require challenging, time-consuming calibration for young children. This limitation has hindered the collection of high-quality visual attention data in naturalistic situations, vital for reflecting real-world infant experiences. A key challenge has been the absence of a suitable wearable eye tracking system specifically designed for infants under two years old.
A New Approach: Eye Tracking for the Youngest Users
To address this challenge, Jérémie Josse, a researcher from the BabyLab – Laboratory of Psychology and NeuroCognition at Grenoble Alpes University, has created a new wearable frame designed for Pupil Labs’ Neon eye tracking module, specifically adapted for infants aged 3 months to toddlers up to 24 months. While Pupil Labs already offers wearable systems for children aged 2 to 8 years (such as the “Crawl walk run” and “All fun and games” frames), no dedicated solution has been available for children under 2. It is exciting to see the research community take initiative and build on Neon’s open geometry to serve this younger population.
The Pupil Labs Neon system was suitable for the project thanks to its lightweight, compact design and modular structure. Its open-source geometry allows researchers to design and build custom frames tailored to their specific needs. Most importantly, it operates without requiring calibration, using a deep learning approach that provides accurate and calibration-free gaze data right out of the box.
This new development aims to combine the precision of eye tracking with the freedom of movement offered by video annotation methods. This enables researchers to gather high-quality visual attention data in natural situations that are more representative of an infant's actual world.
Video demonstrating the usage of Pupil Labs’ Neon eye-tracking module with infants aged 3 to 24 months. Source: Using Pupil Labs' Neon eye-tracking module for infants aged 3 to 24 months.
Designing and Using the Headset
Design: The headset uses a wide flat elastic band, a 3D-printed holder for the Neon module, and includes soft fabric for comfort and a plastic buckle for adjustment.
Fitting Process: Infants typically sit on a parent's lap while an experimenter uses a toy to distract them. The process involves two steps: first, adjusting the empty elastic headband on the infant's head, then quickly attaching the eye tracking module to the front using hook and loop fasteners (Velcro). This two-step method improves acceptance rates compared to attaching the module beforehand. The Neon Companion App's eye camera view helps guide precise module placement.
Testing the Design: Promising Results
Initial studies using the new headset have shown encouraging results in real-world settings:
It achieved a notable acceptance rate of 74% in laboratory conditions with infants aged 3 to 12 months (n=70).
In nurseries, an even higher acceptance was observed, with a 94% acceptance rate for infants aged 4 to 14 months (n=18).
Overall, 56% of the infants tested were successfully included in the study, indicating a solid rate of usable data collection.
Refining the Approach: Future Research Directions
While effective, ongoing efforts are focused on refining the headset’s design and user experience. Development aims to streamline the module placement process for greater ease and consistency, with a second version of the headband already in progress. Work is also dedicated to further optimizing infant comfort and the secure fit of the headset, addressing natural infant curiosity or attempts to interact with the device. Additionally, continuous refinements are set to strengthen the headband's stability to better accommodate typical infant movements without affecting module position.
Overall, this work represents a meaningful advance in infant research, offering a vital tool to study infant visual attention in natural, dynamic environments, while also highlighting how adapting existing technology through open-source methods can expand research possibilities for important populations like infants.
Key Highlights
A new wearable headset for Pupil Labs' Neon eye tracking module has been developed for infants aged 3 to 24 months.
This innovation enables calibration-free, lightweight eye tracking for naturalistic studies of infant visual attention.
The system showed high acceptance rates in both laboratory (74%) and nursery (94%) settings.
It facilitates real-world data collection, offering new insights into infant cognition and development.
Future development is focused on simplifying module placement and enhancing comfort and stability for infant population.
Extend and Develop
Full article: https://hal.science/hal-05007603
Github repository: https://github.com/JerJoss/InfantEyeTrackingHeadband
Lab: BabyLab – Laboratory of Psychology and NeuroCognition Grenoble Alpes University, Grenoble, France
Neon CAD files: https://github.com/pupil-labs/neon-geometry
Documentation on designing your own frames: https://docs.pupil-labs.com/neon/hardware/make-your-own-frame/
The Challenges of Infant Eye Tracking
Studying how infants visually engage with their environment is crucial for understanding early cognitive development. Historically, research has depended on lab-based eye tracking setups or video analysis from head-mounted cameras, which often restrict studies to controlled settings and require challenging, time-consuming calibration for young children. This limitation has hindered the collection of high-quality visual attention data in naturalistic situations, vital for reflecting real-world infant experiences. A key challenge has been the absence of a suitable wearable eye tracking system specifically designed for infants under two years old.
A New Approach: Eye Tracking for the Youngest Users
To address this challenge, Jérémie Josse, a researcher from the BabyLab – Laboratory of Psychology and NeuroCognition at Grenoble Alpes University, has created a new wearable frame designed for Pupil Labs’ Neon eye tracking module, specifically adapted for infants aged 3 months to toddlers up to 24 months. While Pupil Labs already offers wearable systems for children aged 2 to 8 years (such as the “Crawl walk run” and “All fun and games” frames), no dedicated solution has been available for children under 2. It is exciting to see the research community take initiative and build on Neon’s open geometry to serve this younger population.
The Pupil Labs Neon system was suitable for the project thanks to its lightweight, compact design and modular structure. Its open-source geometry allows researchers to design and build custom frames tailored to their specific needs. Most importantly, it operates without requiring calibration, using a deep learning approach that provides accurate and calibration-free gaze data right out of the box.
This new development aims to combine the precision of eye tracking with the freedom of movement offered by video annotation methods. This enables researchers to gather high-quality visual attention data in natural situations that are more representative of an infant's actual world.
Video demonstrating the usage of Pupil Labs’ Neon eye-tracking module with infants aged 3 to 24 months. Source: Using Pupil Labs' Neon eye-tracking module for infants aged 3 to 24 months.
Designing and Using the Headset
Design: The headset uses a wide flat elastic band, a 3D-printed holder for the Neon module, and includes soft fabric for comfort and a plastic buckle for adjustment.
Fitting Process: Infants typically sit on a parent's lap while an experimenter uses a toy to distract them. The process involves two steps: first, adjusting the empty elastic headband on the infant's head, then quickly attaching the eye tracking module to the front using hook and loop fasteners (Velcro). This two-step method improves acceptance rates compared to attaching the module beforehand. The Neon Companion App's eye camera view helps guide precise module placement.
Testing the Design: Promising Results
Initial studies using the new headset have shown encouraging results in real-world settings:
It achieved a notable acceptance rate of 74% in laboratory conditions with infants aged 3 to 12 months (n=70).
In nurseries, an even higher acceptance was observed, with a 94% acceptance rate for infants aged 4 to 14 months (n=18).
Overall, 56% of the infants tested were successfully included in the study, indicating a solid rate of usable data collection.
Refining the Approach: Future Research Directions
While effective, ongoing efforts are focused on refining the headset’s design and user experience. Development aims to streamline the module placement process for greater ease and consistency, with a second version of the headband already in progress. Work is also dedicated to further optimizing infant comfort and the secure fit of the headset, addressing natural infant curiosity or attempts to interact with the device. Additionally, continuous refinements are set to strengthen the headband's stability to better accommodate typical infant movements without affecting module position.
Overall, this work represents a meaningful advance in infant research, offering a vital tool to study infant visual attention in natural, dynamic environments, while also highlighting how adapting existing technology through open-source methods can expand research possibilities for important populations like infants.
Key Highlights
A new wearable headset for Pupil Labs' Neon eye tracking module has been developed for infants aged 3 to 24 months.
This innovation enables calibration-free, lightweight eye tracking for naturalistic studies of infant visual attention.
The system showed high acceptance rates in both laboratory (74%) and nursery (94%) settings.
It facilitates real-world data collection, offering new insights into infant cognition and development.
Future development is focused on simplifying module placement and enhancing comfort and stability for infant population.
Extend and Develop
Full article: https://hal.science/hal-05007603
Github repository: https://github.com/JerJoss/InfantEyeTrackingHeadband
Lab: BabyLab – Laboratory of Psychology and NeuroCognition Grenoble Alpes University, Grenoble, France
Neon CAD files: https://github.com/pupil-labs/neon-geometry
Documentation on designing your own frames: https://docs.pupil-labs.com/neon/hardware/make-your-own-frame/
The Challenges of Infant Eye Tracking
Studying how infants visually engage with their environment is crucial for understanding early cognitive development. Historically, research has depended on lab-based eye tracking setups or video analysis from head-mounted cameras, which often restrict studies to controlled settings and require challenging, time-consuming calibration for young children. This limitation has hindered the collection of high-quality visual attention data in naturalistic situations, vital for reflecting real-world infant experiences. A key challenge has been the absence of a suitable wearable eye tracking system specifically designed for infants under two years old.
A New Approach: Eye Tracking for the Youngest Users
To address this challenge, Jérémie Josse, a researcher from the BabyLab – Laboratory of Psychology and NeuroCognition at Grenoble Alpes University, has created a new wearable frame designed for Pupil Labs’ Neon eye tracking module, specifically adapted for infants aged 3 months to toddlers up to 24 months. While Pupil Labs already offers wearable systems for children aged 2 to 8 years (such as the “Crawl walk run” and “All fun and games” frames), no dedicated solution has been available for children under 2. It is exciting to see the research community take initiative and build on Neon’s open geometry to serve this younger population.
The Pupil Labs Neon system was suitable for the project thanks to its lightweight, compact design and modular structure. Its open-source geometry allows researchers to design and build custom frames tailored to their specific needs. Most importantly, it operates without requiring calibration, using a deep learning approach that provides accurate and calibration-free gaze data right out of the box.
This new development aims to combine the precision of eye tracking with the freedom of movement offered by video annotation methods. This enables researchers to gather high-quality visual attention data in natural situations that are more representative of an infant's actual world.
Video demonstrating the usage of Pupil Labs’ Neon eye-tracking module with infants aged 3 to 24 months. Source: Using Pupil Labs' Neon eye-tracking module for infants aged 3 to 24 months.
Designing and Using the Headset
Design: The headset uses a wide flat elastic band, a 3D-printed holder for the Neon module, and includes soft fabric for comfort and a plastic buckle for adjustment.
Fitting Process: Infants typically sit on a parent's lap while an experimenter uses a toy to distract them. The process involves two steps: first, adjusting the empty elastic headband on the infant's head, then quickly attaching the eye tracking module to the front using hook and loop fasteners (Velcro). This two-step method improves acceptance rates compared to attaching the module beforehand. The Neon Companion App's eye camera view helps guide precise module placement.
Testing the Design: Promising Results
Initial studies using the new headset have shown encouraging results in real-world settings:
It achieved a notable acceptance rate of 74% in laboratory conditions with infants aged 3 to 12 months (n=70).
In nurseries, an even higher acceptance was observed, with a 94% acceptance rate for infants aged 4 to 14 months (n=18).
Overall, 56% of the infants tested were successfully included in the study, indicating a solid rate of usable data collection.
Refining the Approach: Future Research Directions
While effective, ongoing efforts are focused on refining the headset’s design and user experience. Development aims to streamline the module placement process for greater ease and consistency, with a second version of the headband already in progress. Work is also dedicated to further optimizing infant comfort and the secure fit of the headset, addressing natural infant curiosity or attempts to interact with the device. Additionally, continuous refinements are set to strengthen the headband's stability to better accommodate typical infant movements without affecting module position.
Overall, this work represents a meaningful advance in infant research, offering a vital tool to study infant visual attention in natural, dynamic environments, while also highlighting how adapting existing technology through open-source methods can expand research possibilities for important populations like infants.
Key Highlights
A new wearable headset for Pupil Labs' Neon eye tracking module has been developed for infants aged 3 to 24 months.
This innovation enables calibration-free, lightweight eye tracking for naturalistic studies of infant visual attention.
The system showed high acceptance rates in both laboratory (74%) and nursery (94%) settings.
It facilitates real-world data collection, offering new insights into infant cognition and development.
Future development is focused on simplifying module placement and enhancing comfort and stability for infant population.
Extend and Develop
Full article: https://hal.science/hal-05007603
Github repository: https://github.com/JerJoss/InfantEyeTrackingHeadband
Lab: BabyLab – Laboratory of Psychology and NeuroCognition Grenoble Alpes University, Grenoble, France
Neon CAD files: https://github.com/pupil-labs/neon-geometry
Documentation on designing your own frames: https://docs.pupil-labs.com/neon/hardware/make-your-own-frame/