AISO: Annotation of Image Segments using Ontologies

We present AISO, an interactive image segmentation and ontology annotation tool. AISO is designed to give curators of biological specimens the ability to segment an image, such as those produced with digital photography or scanned from printed material, and annotate those segments with ontology terms. We developed this application in response to a need within the plant science community for a streamlined tool that enables the consistent and structured labeling of biological images. To this end, AISO brings together disparate image segmentation, semantic labeling, and data export features found in existing technologies and merges them into a user-friendly, science-focused software package. Given these capabilities, researchers within the biology community are empowered to construct meaningful image data sets drawn from publicly-available online image archives and publications. The interface is specifically designed to streamline the segmentation process in order to reduce the number of required manual corrections. AISO also directly integrates Plant Ontology terms via a lightweight web service, allowing users to select and apply appropriate plant terms to segmented images. The resulting image and ontology label data structure will enable consistent data extraction techniques, efficient database storage, and future semantic inference and active learning functionalities intended to reduce user effort and optimize the annotation process. AISO is a freely available, Java-based desktop application (http://jaiswallab.cgrb.oregonstate.edu/software).

San Diego, CA