Head-mounted cameras have been used in developmental psychology research for more than a decade to provide a rich and comprehensive view of what infants see during their everyday experiences. However, variation between these devices has limited the field’s ability to compare results across studies and across labs. Further, the video data captured by these cameras to date has been relatively low-resolution, limiting how well machine learning algorithms can operate over these rich video data. Here, we provide a well-tested and easily constructed design for a head-mounted camera assembly—the BabyView—developed in collaboration with Daylight Design, LLC., a professional product design firm. The BabyView collects high-resolution video, accelerometer, and gyroscope data from children approximately 6–30 months of age via a GoPro camera custom mounted on a soft child-safety helmet. The BabyView also captures a large, portrait-oriented vertical field-of-view that encompasses both children’s interactions with objects and with their social partners. We detail our protocols for video data management and for handling sensitive data from home environments. We also provide customizable materials for onboarding families with the BabyView. We hope that these materials will encourage the wide adoption of the BabyView, allowing the field to collect high-resolution data that can link children’s everyday environments with their learning outcomes.
This is a preview of subscription content, access via your institution.
Buy single article
Instant access to the full article PDF.
Price excludes VAT (USA)
Tax calculation will be finalised during checkout.
This product and/or service is not affiliated with, endorsed by or in any way associated with GoPro Inc. or its products and services. GoPro, HERO, and their respective logos are trademarks or registered trademarks of GoPro, Inc.
Note that recording at very high resolutions—while permitted by the GoPro camera—more easily drains our smaller, lightweight battery and often causes the GoPro to overheat and automatically shut down. Given that current commercial lightweight batteries are inadequate for powering and maintaining the camera at high resolutions, innovations in lightweight batteries or a custom-designed battery may alleviate these issues and could be attached to the GoPro without redesigning the BabyView camera mount and setup.
Using codebase at https://github.com/tensorflow/tpu/tree/master/models/official/mask_rcnn
For adults and older children, the BabyView camera enclosure is unnecessary, and the easiest possible solution might simply be to use a GoPro 10 or other model and a standard action sports helmet with GoPro mounting; such helmets are commercially available at low cost.
Aslin, R. N. (2009). How infants view natural scenes gathered from a head-mounted camera. Optometry and Vision Science: Official Publication of the American Academy of Optometry, 86(6), 561–565.
Bergelson, E., & Aslin, R. N. (2017). Nature and origins of the lexicon in 6-mo-olds. Proceedings of the National Academy of Sciences, 114(49), 12916–12921.
Borjon, J. I., Abney, D. H., Yu, C., & Smith, L. B. (2021). Head and eyes: Looking behavior in 12-to 24-month-old infants. Journal of Vision, 21(8), 18–18.
Clerkin, E. M., Hart, E., Rehg, J. M., Yu, C., & Smith, L. B. (2017). Real-world visual statistics and infants’ first-learned object names. Philosophical Transactions of the Royal Society B, 372(1711), 20160055.
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009). ImageNet: A large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition, 248–255.
Fausey, C. M., Jayaraman, S., & Smith, L. B. (2016). From faces to hands: Changing visual input in the first two years. Cognition, 152, 101–107.
Feichtenhofer, C., Fan, H., Xiong, B., Girshick, R., & He, K. (2021). A large-scale study on unsupervised spatiotemporal representation learning. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 3299–3309.
Forster, C., Carlone, L., Dellaert, F., & Scaramuzza, D. (2016). On-manifold preintegration for real-time visual–inertial odometry. IEEE Transactions on Robotics, 33(1), 1–21.
Franchak, J. M., Kretch, K. S., & Adolph, K. E. (2017). See and be seen: Infant–caregiver social looking during locomotor free play. Developmental Science, 21(4), e12626.
Franchak, J. M., Kretch, K. S., Soska, K. C., & Adolph, K. E. (2011). Head-mounted eye tracking: A new method to describe infant looking. Child Development, 82(6), 1738–1750.
Grauman, K., Westbury, A., Byrne, E., Chavis, Z., Furnari, A., Girdhar, R., ...,, Liu, X. (2022). Ego4d: Around the world in 3,000 hours of egocentric video. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 18995–19012.
He, K., Gkioxari, G., Dollár, P., & Girshick, R. (2017). Mask R-CNN. Proceedings of the IEEE International Conference on Computer Vision, 2961–2969.
Kretch, K. S., Franchak, J. M., & Adolph, K. E. (2014). Crawling and walking infants see the world differently. Child Development, 85(4), 1503–1518.
Long, B., Kachergis, G., Agrawal, K., & Frank, M. C. (2022a). A longitudinal analysis of the social information in infants’ naturalistic visual experience using automated detections. Developmental Psychology, 58(12), 2211.
Long, B., Sanchez, A., Agrawal, K., Kraus, A. M., & Frank, M. C. (2022b). Automated detections reveal the social information in the changing infant view. Child Development, 93(1), 01–116.
Morgado, P., Vasconcelos, N., & Misra, I. (2021). Audio-visual instance discrimination with cross-modal agreement. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 12475–12486.
Nagarajan, T., & Grauman, K. (2021). Shaping embodied agent behavior with activity-context priors from egocentric video. Advances in Neural Information Processing Systems, 34, 29794–29805.
Orhan, E., Gupta, V., & Lake, B. M. (2020). Self-supervised learning through the eyes of a child. Advances in Neural Information Processing Systems, 33, 9960–9971.
Smith, L. B., Yu, C., Yoshida, H., & Fausey, C. M. (2015). Contributions of head-mounted cameras to studying the visual environments of infants and young children. Journal of Cognition and Development, 16(3), 407–419.
Sugden, N. A., Mohamed-Ali, M. I., & Moulson, M. C. (2014). I spy with my little eye: Typical, daily exposure to faces documented from a first-person infant perspective. Developmental Psychobiology, 56(2), 249–261.
Sullivan, J., Mei, M., Perfors, A., Wojcik, E., & Frank, M. C. (2021). SAYCam: A large, longitudinal audiovisual dataset recorded from the infant’s perspective. Open Mind, 5, 20–29.
Tomar, S. (2006). Converting video formats with FFmpeg. Linux Journal, 2006(146), 10.
Yoshida, H., & Smith, L. B. (2008). What’s in view for toddlers? Using a head camera to study visual experience. Infancy, 13(3), 229–248.
Zhuang, C., Yan, S., Nayebi, A., Schrimpf, M., Frank, M. C., DiCarlo, J. J., & Yamins, D. L. (2021). Unsupervised neural network models of the ventral visual stream. Proceedings of the National Academy of Sciences, 118(3), e2014196118.
This work was funded by a gift from the Schmidt Futures Foundation to M.C.F. and a Stanford Human-Centered AI Hoffman-Yee Research Grant to M.C.F. and D.L.K.Y. and an NIH K99 HD108386 Award to B.L.L.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Safety measures taken during design development.
The head-mounted camera design was developed in concert with a human-centered product design consultancy, Daylight Design. The design consists of a GoPro camera attached to a soft helmet via a custom PLA mount. The safety measures and factors that were considered as part of the design process include: 1. Research into non-toxic 3D printable materials, resulting in the choice of PLA plastic. 2. Use of a consumer soft helmet to serve as the base of the apparatus (Safehead Baby helmets) that was designed to protect infants from harm in the event of a fall (The helmet complies with all relevant safety standards for the U.S. and E.U.) 3. Using corrective helmets for infants to use as a benchmark for an acceptable weight of the entire apparatus (~ 225 g). The full apparatus has been weighed and confirmed to be in a safe weight range. 4. Battery and board components were designed to be separated from direct contact with skin, and the temperature of all parts of the apparatus was confirmed not to exceed 120 degrees F despite extended use [20 min]. 5. Informal pediatrician consultations were conducted, as well as a formal consultation with Paul Fisher (Stanford Pediatric Neurology) where Dr. Fisher was able to handle and evaluate a previous prototype in person. All pediatricians Daylight spoke to expressed that the design as planned should not pose any risks to children. 6. For the final build of the head-mounted camera units, we conducted the safety testing outlined below that aims to verify that the design satisfies the intent (without necessarily using the specific testing equipment) outlined in the testing procedures from 16 CFR § 1500.51 (CPSC guidelines for consumer products for children aged 18 months or less).
Final safety testing protocol:
Developed using CPSC regulation16 CFR § 1500.51 as a guideline:
Drop test: Drop one unit onto a concrete surface covered by 1 / 8-inch (0.3 cm) nominal thickness of type IV vinyl-composition tile ten times, at random orientations, from a height of 4.5 feet (1.37 m). Allow it to come to rest on its own, and examine the unit after each drop.
Conducted 7/7/23_ Notes: All damage to the unit was cosmetic; no components were dislodged.
Bite test: For the protruding surfaces of the battery and camera enclosure (which could afford a mouth getting purchase around them), have an adult bite down with near to their maximum ability over the course of 5 s, and then hold the bite for 10 s.
Conducted 7/7/23__ Notes: Conducted on GoPro (bite mark visible), casing, and battery. No substantive damage.
Tension testing: The helmet will be clamped to a table and held against moving along the sagittal and coronal axis of the helmet. Secure the force gauge to the front face of the camera enclosure so that force will be evenly applied. Over the course of 5 s, pull the force gauge until it reads 10 lbs. of force, and hold for 10 s. Repeat this test, securing the force gauge to the top of the battery enclosure, and to any other projection of the unit that could be grasped with a thumb and forefinger, or teeth.
Conducted 7/7/23__ Notes:
Passed on front of camera.
Directly applying >10 pounds to the interior of the wire enclosure on the 3D printed casing broke off a small part (< 1 cm of PLA tubing) of wire encasing and dislodged the wires from the GoPro connector. Note that this would be difficult to do with fingers from adults/children and would require the wire casing getting stuck on a protruding object. Picture below.
Directly applying >10 pounds of force the remainder of the wire enclosure resulted in it dislodging from the rest of the PLA mounting.
Applied >10 pounds of force to the center of the mounting piece, the bridge that connects the camera to the mount, and to the battery mount. No pieces were dislodged.
No movement on the battery enclosure or battery.
Compression test: Secure the unit so one side of the camera enclosure is flat on the table. Using a force gauge attached to a 1.125-in circular disk, press down on the opposite side of the camera enclosure until the force gauge reads at least 20 lbs. Repeat for the top and bottom of the camera enclosure, and for the side walls of the battery enclosure.
Conducted 7/7/23__ Notes: No issues.
Helmet Strap Test: Holding helmet, have another person pull with a force gauge with hook attached.
Conducted 7/7/23__ Notes: Passed - Strap didn’t break with 60 lbs. 5 lbs. pressure dislodges Velcro strap when hooked on folded portion of strap.
Wire Pulling Test:
a. Conducted 7/7/23__ Notes: Only takes less than a pound of force to disconnect the wires, which are then separate from both the camera and the battery
7. Radio-frequency emissions: We verified via GoPro FCC submissions that radio-frequency (RF) emissions with bluetooth and WiFi functions disabled were compliant with US standards for RF exposure; the maximum power density in the worst case for the lowest level of Bluetooth frequency (BLE) was .000523 mW/cm2 at 20cm. Detailed reports are available at https://osf.io/kwvxu/.
About this article
Cite this article
Long, B., Goodin, S., Kachergis, G. et al. The BabyView camera: Designing a new head-mounted camera to capture children’s early social and visual environments. Behav Res (2023). https://doi.org/10.3758/s13428-023-02206-1