| CPC A01K 11/008 (2013.01) [G06T 7/20 (2013.01)] | 22 Claims |
|
1. A method comprising:
training, by at least one processor, an artificial intelligence model using a library of data associated with poultry, the artificial intelligence model trained based on pre-trained data from other farms before a first time and after the first time training the artificial intelligence model using the pre-trained data from other farms as a first percentage that decreases and using data from a particular location that increases as a second percentage greater than the first percentage over time as data from the particular location increases;
receiving, by the at least one processor, realtime information associated with a plurality of poultry from at least one imaging device at the particular location;
determining, by the at least one processor, at least one benchmark based on the realtime information associated with the plurality of poultry from the at least one imaging device and using the artificial intelligence model by performing higher-bandwidth computing tasks associated with the model by at least one shed computing device and performing lower-bandwidth computing tasks associated with the model by at least one server computing device;
adding, by the at least one processor, the realtime information to the library of data and retraining the artificial intelligence model using the realtime information; and
generating and sending, by the at least one processor, an alert in realtime to a client computing device based on the realtime information associated with the plurality of poultry.
|