Learning to detect salient object in an image based on boolean user input

I am a developer on the Wildlife@Home volunteer computing project and am in the process of researching different supervised machine learning algorithms.

Data for the project is collected by volunteers who classifying wildlife video and the results are grouped for classifying each frame of the video. We have a set of data to determine which frames contain a bird and which do not. The goal is to use this as training data for a machine learning algorithm to predict bird presence in unwatched videos.

Do you know of any previous research in this area or have any suggestions for implementation?