IST Lunch Bunch

Extracting reliable quantitative information from digital images in an automatic
fashion continues to be a difficult task. In many situations classical and
contemporary algorithms only provide partial and sub-optimal results that might
not be sufficient to carry on research studies thus leading practitioners to
rely on manual annotations.  We present our work on collaborative image
segmentation, an online crowdsourcing system where computers, experts, and
non-experts cooperate to produce robust results supporting the research of plant
biologists. We address some of the technical and nontechnical challenges in
building such a system and discuss the potential in employing the vision of
crowds to help solve image processing problems which are still poorly
solved by computers alone.

This is a work in progress in collaboration with Elliot Meyerowitz lab at
Caltech and with Tsang Ing Ren lab at UFPE, Brazil.