Corporate Site
  1. Home
  2. Journal
  3. About
  4. Contact
Corporate Site
Case Study

World's First AI Sorting System for Chilean Sea Urchins: Solving Local Challenges Through Tailored Solutions

READ MORE
Case Study

World’s First AI Sorting System for Chilean Sea Urchins: Solving Local Challenges Through Tailored Solutions

Contents

Back to Index

CHECK POINT

  • AI image recognition streamlines the sea urchin sorting process
  • The system is expected to improve quality consistency and boost sales
  • Effective AI implementation requires optimization within existing constraints

Cost and Risk in Sea Urchin Sorting

While Digital Transformation (DX) accelerates across industries, specific sectors resist technological adoption. Primary industries such as fisheries often continue to rely on legacy tools like fax machines, indicating a significant gap in sector-wide digitalization.

This gap presents substantial transformation opportunities. Marubeni Seafoods, a global marine products distributor, has pioneered digital transformation initiatives in the fisheries industry. A standout project is their implementation of AI-based image recognition technology in their Chilean sea urchin operations.

Chile currently accounts for approximately 40% of global sea urchin harvests, with Chilean products representing 95% of frozen sea urchins imported to Japan. Despite their market dominance, the sorting process between harvesting and shipping has remained a critical operational challenge.

Sea urchins are graded from A1 to B2 primarily based on color characteristics. A single grade difference can alter a 100g package’s value by hundreds of yen (several dollars), making accurate classification essential for profitability. Additionally, color inconsistency within 100g packages increases the risk of customer complaints. Traditional sorting relied on visual inspection by workers using PANTONE color standards as a reference, but human judgment introduced inevitable inconsistencies.

Even with dedicated staff, evaluating countless sea urchins with subtle color variations leads to standard drift and fatigue-induced errors. These inconsistencies significantly increased quality control costs in Japan, sometimes requiring multiple staff members to spend days manually inspecting individual 100g packages—an unsustainable process.

Quantifying Color with AI Image Recognition

Marubeni Seafoods addressed this challenge by implementing AI image recognition technology.

The initiative began in 2020 with basic camera systems to identify sea urchin coloration. During COVID-19 restrictions, the team conducted preliminary testing with images collected in Japan. By 2022, on-site reconnaissance and field testing commenced, allowing for solutions more precisely tailored to actual operational conditions.

Initially, the development team explored deep-learning approaches using sea urchin image datasets. However, they rejected “black box” AI systems that would obscure decision-making rationales and potentially reduce acceptance among Chilean workers. Instead, they developed a transparent system that extracts sea urchins from camera imagery, analyzes their RGB color components, and classifies them numerically based on predefined grade thresholds—creating an objective standard all stakeholders understand.

The implementation extended beyond algorithm development. Custom hardware solutions are required to operate reliably in the cold, humid Chilean processing facilities. Additionally, the interface needed extensive refinement to ensure usability by staff without specialized technical training. In 2023, Marubeni partnered with specialized AI hardware ventures to enhance system stability and conduct extensive on-site testing. After iterative improvements to both algorithms and user experience design, the base system was completed and deployed as a functional application in 2024, with factory implementation currently underway.

This system’s deployment is early, with comprehensive impact assessments planned. The team recognizes that technical functionality alone is insufficient—success requires independent operation by local staff. Marubeni provides ongoing support while developing operational frameworks, with plans to expand implementation across additional processing facilities.

Technology Meets Reality: The Pragmatic Approach

The image recognition system developed for sea urchin processing has potential applications across numerous sectors. Color assessment represents a critical quality indicator throughout the seafood industry, suggesting natural expansion opportunities for other marine products.

Terms like “AI utilization” and “digital transformation” often evoke images of cutting-edge technological solutions. However, the core technology implemented here—RGB color decomposition and analysis—has existed for decades and remains relatively straightforward from a technical perspective.

This highlights an important reality: advanced technology alone doesn’t solve business problems. Effective problem-solving and business enhancement depend more on appropriate technology integration than technological sophistication.

While specific challenges require advanced technologies like autonomous vehicles or robotics, practical business environments operate under numerous constraints, including budgetary limitations. Creating business impact, therefore, requires maximizing outcomes within existing constraints. This approach exemplifies how general trading companies like Marubeni, rather than specialized tech firms, can effectively leverage digital technologies.

By developing solutions that address local challenges through sincere engagement with operational realities, Marubeni demonstrates a pragmatic approach to digital transformation that avoids technology-for-technology-sake thinking. This methodology, focused on practical outcomes rather than technological showcasing, provides a model for digital initiatives across diverse sectors.

Back to Index

Related Journal

Get in Touch