Sign up for our daily Newsletter and stay up to date with all the latest news!

Subscribe I am already a subscriber

You are using software which is blocking our advertisements (adblocker).

As we provide the news for free, we are relying on revenues from our banners. So please disable your adblocker and reload the page to continue using this site.

Click here for a guide on disabling your adblocker.

Sign up for our daily Newsletter and stay up to date with all the latest news!

Subscribe I am already a subscriber

“Using AI to grade and sort your cannabis ensures consistent high quality”

“We developed an optical grading system that uses cameras and AI to grade cannabis at high speed. It will look for deficiencies or specific criteria and separate those out automatically,” explains Jay Evans, CEO of Keirton. One of their newest solutions, Marvel, will be released later this year and is developed to help cannabis companies in today’s challenging climate. “The main premise is to ensure that the producer gets a consistent product to their customer every time. Inconsistency will give your brand a bad reputation, which this solution helps to eliminate.”

How does it work?
Jay explains that a conveyor belt transports the flower, which goes through an array of cameras. “These cameras take images at a really high speed and from all angles. Based on the images, it decides whether the pieces are good or bad. Bad pieces will be separated with a puff of air.”

The system can identify many important things, such as mold, too much stem, too much leaf, crow’s feet, or poor color. The producer can select which criteria are important for them and can choose different parameters within each criteria. “Perhaps stem is your main priority, and only under a quarter of an inch is okay. Or perhaps mold is of the most importance to you. One of our customers is a greenhouse grower who struggles with mold in the summer. Trying to find the odd moldy piece is very time-consuming and can sometimes still be missed. The Marvel will kick it out automatically, at high speed.”

Rejection Mode: High-resolution images of both sides of every flower 

Similar solutions are used in many other agricultural industries, Jay says. “Yet there is nothing like this in cannabis yet, and it was very challenging to develop. Most fruits are relatively consistent in shape and color, so any issues are obvious to see. Cannabis, on the other hand, can come in so many shapes and colors. With a wide variety of deficiencies, it becomes very challenging to grade. We’ve had to take millions of cannabis images as the data set. We’re continuing to take more and more images, perfecting the model.”

Analytic Mode: Know exactly what is in the batch

Saving reputations
According to Jay, inconsistency is a major issue among cannabis companies. “Occasionally, a bad piece of flower will make it through to the consumer. It might have powdery mildew on it, or it might not be trimmed well, for example. When a consumer gets that, it’s not a good look for the brand. Consumers want a consistent experience. With cannabis being so inconsistent, you’re often relying on humans to grade it and select what is best. But then there is a possible human error that could come into it. People get tired after looking through flower for hours. Some may not be as diligent as others, or someone might be having a bad day. Moreover, there is a bias as to what is good or bad. That is where the Marvel becomes very useful, as it provides consistency without any bias.”

The company has already soft-launched the system for a select group of customers, and the feedback has been very positive. “Some are saying it’s made a dramatic difference in their bottom line. Some companies would normally sample an entire lot and if they didn’t meet a certain spec, they would have to downgrade that entire lot to a lower-level brand. Now, they’re able to pull out the bad pieces and keep the batch for their high-level brand. This way, the value it creates for the company is tremendous.”

Another company was able to save labor and rearrange their staff to other positions, Jay says. “The company had 10 people on their quality control line, looking through flower. Since using the Marvel, they were able to move that down to just 2 people. The other 8 people now work in other parts of the facility, which is great as they had a hard time finding staff. Saving labor is another major benefit of the system, but the biggest benefits are definitely leveling up the brand quality and saving reputations.”

For more information:
Twister Technologies