At a glance
Imagine if each coffee bean came with a label communicating its level of quality or a specific quality trait, e.g. level of fermentation. This would make it very simple for coffee manufacturers to select beans by quality or to create specific blends based on specific traits that affect taste and flavor. In a manner of speaking, this is exactly what Seed-X GeNeeTM technology does. Its AI-powered analysis enables quality scoring of beans using visual data rather than human taste data, and moreover, on a bean-by-bean basis.
In this paper, we will demonstrate how, for the first time ever, Seed-X AI powered technology enables instantaneous and simple separation of high quality, properly fermented Arabica coffee beans from poor quality, over-fermented coffee beans, both coming from a single non-uniform coffee bean bag. By doing so, Seed-X technology opens up a world of coffee creation possibilities.
The Context & Problem
Coffee manufacturers are constantly looking for ways to improve the efficiency and efficacy of their QA process and the quality of their coffee blends. Today, coffee roasters and manufacturers rely heavily on human testers to determine the quality of different bags. The process involves selecting samples of beans from various bags, roasting them, and then taste-testing them using highly qualified human testers.
While this time-tested approach has become a convention in the coffee industry, it still holds noticeable drawbacks.
- Need for professional human panel to perform the analysis. Today, coffee bean bags with undetermined quality are being advanced through the supply chain, processed and then shipped across the world to roasters and manufacturers who define coffee bean bag quality after investing considerable time, effort and resources. If quality is found to be insufficient, coffee bean bags are sent back to the source.
- Limited sampling size. Since coffee testing is performed on a small portion of the bean bag, there is an increased chance of receiving “muddy” or non-representative results.
- Lack of a method for pre-sorting coffee bean bags by quality. Current QA processes are conducted per bean bag and do not attempt to sift through the bean bag. As to date, there is no way to classify the beans by quality attributes bean-by-bean, in a non-destructive manner.
The Seed-X Solution
Computer vision-based, instantaneous, and effortless separation of Arabica coffee beans by fermentation level
Different parameters affect coffee bean quality, such as the bean production conditions, post-harvest processing and storage conditions, which influence bean flavor and acidity, fermentation level and fungal contamination level, respectively. Over-fermentation during coffee bean processing is detrimental to taste and diminishes bean quality. If a coffee bean bag contains a distinguishable proportion of over-fermented beans, the whole bag is likely be discarded.
Seed-X GeNee™ technology has proven successful in classifying a bag of fermented Arabica coffee beans. Through the application of its AI-powered imaging and phenotype analysis, it bypasses the need for conventional QA (i.e., taste-testing) on the coffee manufacturer’s end, facilitating a more efficacious and sustainable supply chain.
How Seed-X GeNeeTM Performed QA Analysis
Step 1. Bean imaging by GeNee™ Detect. Arabica coffee bean bags were supplied by a well-known coffee manufacturer. Seed-X GeNeeTM Detect captured images from thousands of beans coming from 4 properly fermented and 4 over-fermented Arabica coffee bean bags.
Step 2. Building GeNeeTM Envision Classifier. The Seed-X GeNeeTM classifier was trained to distinguish between the different coffee bean bags. Relatedness between the different bags was visualized by a scatter plot, where each dot represents a single coffee bean and each bag is assigned with a distinctive color. The closer the dots are, the greater the relatedness between beans.
Step 3. Smart separation by coffee bean fermentation level. A nineth sample, belonging to an extremely non-uniform coffee bean bag was sorted based on bean relatedness to either one of the two fermentation groups, resulting in two assemblages of putative properly fermented beans and putative over-fermented beans.
Step 4. Taste Testing. Following the Seed-X analysis, the two assemblages were taste-tested by the coffee manufacturer. Taste-tests affirmed the composition of properly- and over-fermented beans and the separation made by GeNeeTM technology.
Quality Analysis Process
The Bottom Line
Seed-X’s AI-powered GeNee™ technology will allow coffee bean producers to make more effective bean selection decisions before advancing the bean bag down the supply chain. Here we demonstrate an automated, instantaneous, and effortless solution to assure proper fermentation in the coffee bean bag before dispatching the bag to roasters and coffee manufacturers. This methodology might be applied to identify additional quality attributes in coffee beans – acidity, Rio flavor and traceability – all of which already yield positive indications in our ongoing experiments.
Future Applications
In the future, Seed-X plans to add coffee quality detection to its GeNee™ Seed Sorter. This will further allow coffee bean producers to sort beans according to specific characteristics and quality types, and to preserve beans that would otherwise be rejected or discarded.
Thinking even further ahead, Seed-X opens up the possibility for coffee manufacturers to create ultra-premium quality coffee blends or to create limitless unique customized blends.
Today, coffee manufacturers can create blends based on the inclusive quality profile of an entire bag (which, as demonstrated, contains quality variations), since they do not have the possibility to sift through bags based on the characteristics of individual beans. Seed-X is inviting coffee manufacturers to imagine creating coffee blends by creating new and highly distinct or customized blends by selecting on a bean-by-bean, rather than a bag-by-bag basis. This would motivate coffee manufacturers to create new, more accurate flavor profiles by ‘curating’ individual beans featuring specific traits.
Soon coffee makers will be able to use Seed-X AI-powered technology to ensure only the best quality coffee beans end up in our morning cup of coffee, and to launch new highly- tailored coffee blends.
To learn how Seed-X can add value to your coffee business, please contact:
info@seed-x.com; or visit us at: https://www.seed-x.com