AI Magazine - Winter 2018

presented work on combining human and machine intelligence to describe images to people with visual impairments. Walter Lasecki (University of Michigan) discussed projects using real-time crowdsourcing to rapidly and scalably generate training data for computer vision systems.

One major theme of the workshop papers was building systems that leverage the visual capabilities of crowdsourced human workers. Abdullah Alshaibani and colleagues at Purdue University presented InFocus, the workshop’s Best Paper award winner. This system enables untrusted workers to redact potentially sensitive content from imagery. Kyung Je Jo and colleagues at KAIST presented Exprgram, the workshop’s Best Paper Runner-Up award winner. This paper introduced a crowd work-flow that supports language learning while annotating and searching videos. Ground Truth, a system by Rachel Kohler and colleagues at Virginia Tech, combines expert investigators and novice crowds to identify the precise geographic location where images and videos were created.

Another major theme of the papers was creating
synergies between crowdsourced human visual
analysis and computer vision techniques. Steven
Gutstein and colleagues from the US Army Research
Laboratory presented a system that integrates a
brain-computer interface with computer vision tech-
niques to support rapid triage of images. Divya
Ramesh and colleagues from CloudSight presented
an approach for real-time captioning of images by
combining crowdsourcing and computer vision.

A third theme of the workshop papers was improving methods for aggregating results from crowdsourced image analysis. Jean Song and colleagues at the University of Michigan presented research showing that tool diversity can improve aggregate crowd performance on image segmentation tasks. Anupar-na Banerjee and colleagues at the University of Texas, Austin presented an analysis of ways that crowd workers disagree in visual question-answering tasks.

The workshop also featured a poster session and break-out groups. Participants used a bottom-up approach to identify topical clusters of common research interests and open problems. These clusters included real-time crowdsourcing, worker abilities, applications (both to computer vision and in general), and crowdsourcing ethics.

Danna Gurari (UT Austin), Kurt Luther (Virginia Tech), Genevieve Patterson (Brown University and Microsoft Research New England), and Steve Branson (Caltech) served as cochairs of this workshop. The steering committee comprised James Hays (Georgia Tech), Pietro Perona (Caltech), and Serge Belongie (Cornell Tech). Some of the papers are available on the workshop website, groupsight.github.io.