Article Preview
TopIntroduction
Human limitations associated with the study of underwater life have made it difficult to study marine species. However, with the continuous growth of technology, marine animal data collecting has become more accessible, presenting growing potential for knowledge acquirement regarding these species. Such knowledge is important in a variety of contexts, both for scientific and commercial purposes. Currently, significant investment has been made in tagging and tracking technology. Resulting data, if properly analyzed, may help to understand intrinsic and external factors that may influence marine species’ behavior, leading to a better overall understanding of the ecosystems in which such animals are inserted. Furthermore, adequate analysis of this information may reveal a number of patterns relative to different species, aiding in the process of understanding the behavior of these animals. Nevertheless, telemetry data gathered through animal tagging presents several challenges that make it hard to analyze. Not only does it often consist of large amounts of information, but it is also frequently found in a textual form, difficult to read and understand. Consequently, deriving knowledge from this information is often a long and extremely demanding process. For instance: (i) Does Oceania in fact have a larger number of recorded Clownfish occurrences when compared to the rest of the world? (ii) Which species is more often recorded in colder months? (iii) Has the Surgeonfish inhabited deeper ocean areas over time?
One way to overcome this challenge is through the use of Information Visualization (InfoVis), which alleviates cognitive load associated with data interpretation (Munzner, 2014). A meaningful visualization will in fact allow the representation of information in a way that highlights relevant information while providing the means for the user to explore and find important patterns.
Taking the aforementioned limitations into account, the main goal of our work was to present marine species’ information in an integrated manner while providing users with new means of learning about fish species and at the same time interacting with such data. To meet this goal we created FishEye, a visualization which presents marine species’ information in an integrated manner while providing users with new means of learning about fish species and at the same time interacting with such data. Furthermore, it allows the comparison of two species simultaneously across all existing information domains, enhancing the identification of common behaviors, as also, establishing relations about such species.
FishEye consists of a dashboard with multiple, interconnected views providing complementary perspectives on the presented information. These views change dynamically with user input, highlighting relevant contextual information and informing the remaining idioms of changes that are taking place.
An iterative and incremental paradigm was followed during development. An early expert evaluation took place with a set of heuristics suited for Information Visualization (Forsell, 2010) which, while pointing directions for the development of our solution, rendered it valid for further development and subsequent testing. Furthermore, to understand how, in real-life context, users react and experience FishEye, quantitative usability metrics were gathered though usability tests.
In order to allow users to identify fish species to further learn about them, a proof-of-concept species’ recognition module was integrated within FishEye. It consists of extracting features about the provided species and then using classification methods to successfully recognize it. To proper validate the recognition performance, an evaluation of this module was carried out which showed FishEye’s potential to efficiently recognize species.
This document is organized as follows. We introduce relevant related work that situates our study on both recognition and information visualization on animal species. We then present and detail our solution, followed by a description of its evaluation. We then draw the main conclusions deriving from our work and a set of guidelines for further improvements on our marine species’ visualization.