AI training for bighorn
Artificial Intelligence (AI) can be a tool for conservation, but just like any other model in ecology, it needs to be properly trained and tested. A number of programs use AI to identify wildlife species photographed by trail cameras. So far, the results show that AI can only get us so far and species ID'ed by a computer model still need to be checked by a human researcher. Can we improve bighorn ID by training these models with better data instead of just more data?
Finding a balance - generalized vs specialized
Can AI learn from its mistakes?When AI fails, we want to know where and why. Just like people, AI's failures may provide lessons to help us improve predictions faster
We plan to:
|
|