Introduction
Collecting and analyzing images of local fauna is crucial to the success of wildlife management and conservation efforts.
State-of-the-art motion-activated cameras enable wildlife professionals and enthusiasts to photograph animals regardless
of date, time, or weather. In turn, the images can be transferred to a computer for visualization, analysis, and insights into
population health and behavior.
Of the many species commonly photographed in North America, whitetail bucks (Odocoileus virginianus) are particularly
captivating due to their majestic morphology and changes in their appearance as they mature. Nonetheless, accurately
predicting the age of a whitetail buck can prove difficult. To predict the age of living bucks, many hunters and enthusiasts
turn to ”aging on the hoof” (AOTH). Found in literature as early as 1978, this method attempts to predict the whitetail’s
age based on the deer’s location, photograph date, and body traits such as the chest, stomach, neck, legs, and antlers.
Prior AOTH research
Previous research efforts have been performed on captive, known-age deer. Flinn’s 2010 study, for instance, analyzed
64 morphometric ratios of living, wild white-tailed deer across three states: Mississippi, Louisiana, and Texas. The study
resulted in two age prediction models, one simple and one complex. When applied to individual age classes, the simple
model achieved 33% accuracy for pre-breeding periods and 40% accuracy for post-breeding periods. The complex model,
on the other hand, achieved 53% accuracy for pre-breeding periods, and 63% accuracy for post-breeding periods.
Another study in 2013 by Gee et al. found that wildlife enthusiasts and professionals achieved a 36% accuracy rate
when aging deer on the hoof, although individual scores widely varied, ranging from 16% to 56%. Furthermore, prediction accuracy was found to decline as the age of the buck increased. For instance, wildlife professionals performed better at
predicting the age of yearling bucks (1.5 years) than older whitetail bucks (ex. 4.5 years). Despite the relatively low
prediction accuracies, the same wildlife professionals believed a minimum accuracy of 70% is needed to be useful in
wildlife management decisions, and a minimum accuracy of 80% is needed for research applications – standards that
neither morphometric models nor human assessors have achieved.
Professional AOTH
The ages of whitetail bucks can be professionally estimated in many different ways. The first is by contributing trail
camera images or video to established organizations like the National Deer Association (NDA), in which a panel of
wildlife biologists and experienced professionals reach a mutual consensus over each photo (Lindsay Thomas, personal
communication, May 1, 2025). To be considered, the deer in the contributed photograph must meet several requirements:
it must be broadside, standing relatively still, have its head up, the image must be well lit and not blurry, it must have its
entire body contained in the image frame, it must be photographed in pre-rut or during rut, and there can be no velvet on
the buck’s antlers.
Even with these requirements in place, a large request volume remains, overwhelmimg wildlife experts and making it
difficult to give equal attention to each submission. On the other hand, if the buck is deceased its jawbone and teeth can
be removed and submitted to a laboratory for further age analysis. Although more scientific than AOTH, this
process can still prove cumbersome and is not always feasible depending on the state of the deer’s body.
To alleviate the workload on wildlife professionals and enhance awareness within the whitetail community, many organizations provide tutorials on best AOTH practices. In each tutorial, highly trained professionals discuss general body
characteristics and aging techniques across different age classes, often providing clear examples to help clarify the message. Over time, the growing database of information, tutorials, and images helps viewers develop an intuition for aging
whitetails that can be applied in their own experiences.
Computer Vision in White-tailed Deer Aging
Much in the same way that humans learn to predict age based on a deer’s body shape, Machine Learning (ML) and
Computer Vision (CV) can be leveraged to generate age prediction models based on the same database of trail camera
imagery and pre-determined ages. Models like Convolutional Neural Networks (CNN) automatically detect and extract
features within trail camera images, removing the manual labor of feature identification exemplified in Flinn’s 2010 thesis.
This study reports the first known use of computer vision and CNNs in predicting the age of whitetail bucks from trail
camera imagery, offering wildlife experts, hunters, and enthusiasts a new tool in wildlife management and conservation
efforts.
Conclusion
AOTH has deep roots in the outdoor community, and is a well established practice inside and outside academic circles. But with the dawn of machine learning and computer vision, new technologies have the potential to overhaul and improve the field of deer herd management.