WWF Deutschland

Automated reading to assess the sustainability of seafood

Description of the company

At the marine department of WWF Germany (located in Hamburg), we have different units: International marine conservation projects, flagship species conservation (sharks, wales,…), national and international politics related to ocean protection, plastics in our oceans and market work on seafood. This project will contribute to the team of market work on seafood, where we coordinate our seafood guide application and all information stored in it. Seafood guides are tools to help consumers buy sustainable fish.

Situation

Seafood guides work in the form of a traffic-light scoring system: from green (sustainable product) to red (unsustainable product). Consumers should find on the package of seafood products they buy in stores (ex. supermarkets) certain relevant information in order to find the matching score in the seafood guide. For instance, a consumer buys some frozen shrimps in the supermarkets, and should find on the package: the Latin name of the shrimp species, where it was farmed or fished, and how it was farmed or fished (exact fishing gear or farming method). Only then the matching score (if existing) can be related to the seafood guide.

Problem

For consumers to find out these specific information on package - like Latin name, region and gear/farming method – is time-consuming and not always easy to detect (sometimes written in small letters, or just somewhere in the middle of thousands of other information). This might dissuade consumers to further use the seafood guide. A new technology is needed to ease the task of consumers.

Aims of the project

In this project, we would like to have a new tool being developed for consumers using their smartphones in supermarkets. With help of their smartphone’s camera, consumers could scan the package of seafood products so that the camera (and technology behind it) would automatically detect the needed information (Latin name, region, gear or farming method).

To achieve this, the programmed software should be able (or be trained) to detect these information automatically, and connect them through the API of our own database, where we have stored all our data. Matching data should be tried to be found, and if the case, the traffic light score should appear on the consumer’s smartphone.

Ideally, this new tool should be an add-on of the existing seafood guide apps. So a tool that can easily be added in existing apps.

Some aspects to consider: the information found on package is not always of the same granularity as the one found in our database. Example: a large region is given on package (ex. FAO 27), but we do have 3 assessments for subregions within FAO 27. Or a fishing gear is mentioned on package, ex. trawlers, but our data differentiate midwater trawler from bottom trawlers. There is a way, dependent on fish species, to make those data match but it needs to be coded (rules should be given, for instance Hering is never caught with bottom trawlers, so if noted trawler on package, it is in fact midwater trawlers.

Target group (students)

The target group should have experience with Optical Character Recognition (OCR) technologies (or similar) in order make the camera read written information on package (after a picture was made, or even better with the live camera). As the seafood guides have mostly been developed with Wordpress, it would be an advantage for those who know Wordpress in case the new tool would be integrated in the app. Any other coding skills is welcome.

Dates
Please save these dates: Fishing for Experience Termine

Registration
You can apply for Fishing for Experience online.