In addressing the urgent challenge of illegal poaching threatening elephants in the Massai region of Kenya, I designed and delivered a B2C dashboard to match unknown elephants with the existing database in 8 weeks.

The dashboard functions as a robust tool, streamlining the application of the SEEK algorithm (developed by CalTech researchers) and incorporating front-end design and development. This integration not only reduces the time needed for matching unidentified elephants but also enhances the capability to effectively categorize distinct matches based on algorithmic data.

To craft the interface, our initial focus was on deciphering how the SEEK algorithm assesses and compares an unknown elephant against the existing database.

Every elephant possesses distinctive ears, with no identical pairs. Each elephant is assigned a CurvRank Score, determined through a numerical analysis of the variations in tears, bumps, and curvature within their ears. Elephants are subsequently ranked based on their CurvRank Score, and newly identified elephants without matches are allocated a unique SEEK code, denoted by a combination of letters and numbers. This code is contingent on factors additional factors such as sex, anticipated age, and tusk characteristics.

According to the SEEK Algorithm Paper, the list of ranked elephants is limited to the top 15. While the majority of elephants typically fall within this top subset, human intervention remains necessary to review various matches and ultimately determine the accuracy of the elephant identification.

Current Interface:

Lacks Elephant Comparison: displaying only the database elephant rank without providing a direct visual link to the unidentified elephant and its matches in the database.

Inefficient: shows only one picture per elephant, despite access to more images, limiting the visual information crucial for rangers engaged in manual identification.

SEEK Codes: challenging to interpret, adds complexity, especially when dealing with unknown information and may overwhelm users.

We identified common threads of valuable features that hold significant potential for implementation including swiping through matches, implementing manual feedback systems, and enhancing information architecture—such as enlarging images—will contribute to a more user-friendly and engaging interface.

*Constraint: pre-existing + immutable visual design elements on the website, our primary emphasis lies in optimizing the usability and functionality of the features, refining the user experience and ensuring that every feature performs seamlessly to meet the conservation objectives effectively.

Unidentified Elephant Column (static) vs. Possible Match Column (dynamic)

Right/Left Arrows Browse Selected Elephant Photos

Up/Down Arrows Browse Possible Matches

Gallery View of Selected Elephant Photos

Seek Code: RED- differences; BLACK- similarities

*Restricted Admin Only Site (unable to share live view)

Using JavaScript, Django, HTML/CSS, the dashboard was successfully programming and integrated into the ElephantBook website.