Designing the Future: How Mission Transformed Noun Project's Visual Discovery Through AI-Powered Style Recognition

Noun Project, the world’s most extensive and diverse collection of icons and mission-driven stock photos, partnered with Mission to solve the critical challenge of style-based visual discovery. Through innovative machine learning and computer vision techniques leveraging AWS services, Mission enabled users to efficiently find style-consistent icon sets. The solution transformed user experience, reduced search friction, and reinforced Noun Project's position as the premier destination for high-quality visual assets.

Noun Project is building a global visual language that unites us — a language that allows quick and easy communication no matter who you are or where you are. The platform offers over 8 million icons and photos,serving a global community of designers, educators, marketers, and organizations with a human-first approach where every icon is created by independent artists and curated by experts.

Noun Project's vast collection, contributed by thousands of artists, offered unparalleled visual options but also created discovery challenges due to its volume. The platform's search tools primarily supported discovery by subject matter, not by style, forcing users to manually sift through thousands of icons to assemble cohesive sets.

Noun Project's Chief Product Officer first discovered Mission at an AWS conference, where their AWS account manager confirmed Mission as an official AWS partner, which provided initial confidence. Noun Project was then drawn to Mission's prior work with AI-assisted tools that helped users build photo albums, demonstrating engineering expertise in image interpretation and familiarity with design-oriented tech tools. When discussing the unique challenges of illustrative style recognition, Mission's engineers were engaged and excited with an outpouring of ideas. Their technical expertise, combined with collaborative working methods, was particularly appealing. Mission worked in partnership with Noun Project's AI engineer and product team, discussing, scoping, and assigning work together under excellent project management, allowing most resources to focus on actual engineering and product development rather than overhead.

Noun Project chose AWS because it provided the ability to instantly serve a global audience with speed and reliability. As a platform hosting millions of creative assets accessed worldwide, it was important to have infrastructure that could scale while maintaining consistent performance. AWS was trusted to ensure the community had fast, dependable access to their library of visual resources.

Mission began by engaging with Noun Project's teams to curate sample sets and establish foundational style categories. Initial applications using generative AI revealed that scaling to capture nuanced qualities of millions of graphic designs required machine learning. Mission pivoted to an ML-led approach, extracting statistical features from icons such as line width and corner radius using computer vision techniques. These objective measurements became the basis for clustering icons by style through multiple rounds of testing, refinement, and validation with design experts.

The scaling strategy was phased and data-driven, starting with small, curated icon sets to validate style definitions, then expanding to larger, randomized groups to stress-test algorithms for accuracy and performance. Each iteration incorporated feedback from technical and creative stakeholders, culminating in a robust system capable of classifying and searching millions of icons in production. The serverless architecture ensured infrastructure costs remained manageable as the platform grew, providing ongoing efficiency through optimal resource utilization.

The transformed user experience delivered dramatic improvements in search and discovery capabilities. Noun Project implemented several new features leveraging Mission's work: "Show Similar" executes searches on visual characteristics and subject matter, "Find Matching Icons" allows users to find different icons matching a primary icon, and filtering user search by style and line weight. These features turned what was once laborious into an engaging, exploratory journey allowing users to focus on creativity rather than curation.

The "Show Similar" feature on Noun Project's website enables search by visual characteristics as well as subject matter.

Users on Noun Project's website can now filter by line weight, icon style, and other artistic differences.

Noun Project user testimonials:

• “I have been using Noun Project since 2014 and have been wanting this feature for literal years. Congratulations on the launch, this is a game-changer!”
• “WOW this is a brilliant feature - I have often wished for this. Thank you! It’s awesome!”
• “I’m really excited by the “Show Similar Icons” and “Search for icons with a similar style” features you’ve added to the web app.”
• “For designers who care about consistency and spend too much time hunting for the perfect icon, this is such a thoughtful touch.”

User feedback has been overwhelmingly positive, with testimonials highlighting newfound ease in creating consistent visual narratives and significant time savings. Social media posts expressed delight at long-wished-for features, with users calling it "a game-changer" and "brilliant." The platform observed significant decreases in dwell time on search pages, indicating users find needed icons much faster than before. This efficiency gain translates to higher engagement and retention among creative professionals. Increased usage of search tools demonstrates users leveraging new capabilities to build cohesive icon sets for presentations, apps, and branding materials. The solution reinforced Noun Project's reputation as an innovator, attracting new users and deepening community engagement while opening new acquisition opportunities.