Newby here,
I sent an email to Peter and received an out of office from him until June 27. I am not sure if AddaxAI can be modified or this is just a custom model that I can possibly
develop with some guidance. See my email to Peter below. If this is not the correct forum for this type of question, please advise.
SSBNJOHN
Peter,
Impressed with your AI, but have just started to determine if it will help with our challenges. I currently help manage/harvest White Tail Deer on a ranch in Texas close to Laredo, TX. Google Maps on “Las Tiendas, TX” if interested. We typically do a 4-6 week camera survey on 11 different location/feed pens that are used to concentrate the deer population. I am responsible for the game camera setup, deployment, and potentially some voodoo to derive population estimates for Texas biologist.
The ranch just purchased new camera’s, so still in the learning curve as far as settings, mounting position, etc. In previous years, we have “poorly” digested/analized ~10,000 pictures from these locations to try and determine/estimate total unique deer. We are convinced our estimates are grossly overstated. We also recently attempted to have a night time infrared survey that was supposed to yield population (98%) accurrate with breakout on Male/Buck, Female/Doe, and Fawn/Juvenile less than 12 months old. The ambient temperture(ground/trees) was so warm when attempted, the survey was not successful enough and is currently being rescheduled.
I need another way to analyze than “ocular”. I am a retired IT Director and have some experience knowledge with vision systems used in manufacturing or automated packaging/labeling.
Just some “sort of facts” about White Tail deer. Each Buck has a unique set of antlers each year. Typically increasing in mass, points, and length, until about age 7-8 and then starts to decline as their teeth wear down and they can not consume enough protein to support optimum growth. Antlers vs horns are shed each Spring and regrown through out April~August. During the growth period, the antlers are covered in “velvet” or a fine hair like substance allowing the horns to grow and extend. The antler configuration and size is determined by genetics of both parents and food available on the range during these months, as well as how bad/good the range the previous year was. Antlers are only grown with excess resources the deer does not need for body growth/nourishment.
Obviously, with our game pictures, the camera location and field of view is fixed once the survey is started. We have it 25-30 feet away from the food source, so it covers the most frequented spot. A solitary buck can be standing in any of 360 degrees to the food and have its head in just about any orientation but rarely upside down.
So the real question is: how difficult would it be to try to uniquely identify and count each of the unique bucks in group of pictures quantity 1500 and preferably ~12000?
Because of the pens(camera locations) proximity to each other, some of the bucks can frequent more than one pen in a given day/night.
I can provide more pictures if you would like or answer any questions you might have. The largest antler set in each of the pictures is from two different pens/cameras.
Regards,