The goal of LabelMe is to provide an online annotation tool to build image databases for computer vision research. Furthermore, LabelMe also lets you train and execute object detectors from the image annotations.
The LabelMe App allows taking pictures and tagging objects with bounding boxes. Users can upload the images and their annotations to the LabelMe web server where other people will be able to access them.
LabelMe also allows using the image annotations to create and train objects detectors. After the training, the detector will be able to recognize objects in real time on the video output of the iOS device.
The goal is to use those annotations for research in object recognition.