Traditionally, this kind of analysis required hundreds of hours of manual work. Our goal was to reduce that burden without asking users to trust a black box. The tool lets ecologists search a map for images that visually match a few examples they select. They then refine and approve the results before adding them to the database—keeping them in control.
My work focused on interface design and front-end development, especially the refinement gallery. This feature lets users scrub through results with precision and confidence, and it quickly became central to the workflow. Later versions introduced a more compact layout, streamlined controls, and a trimming tool that made it easier to explore and fine-tune matches.
Through continuous iteration and close collaboration with our ecology team, Smart Search became more than a feature—it became part of their daily toolkit. With it, we saw up to a 77% reduction in time-to-label, making ecological insights faster to generate and more enjoyable to produce.
