First Fresh NZ, a major citrus supplier to the domestic market in New Zealand, has taken the global lead as the first citrus enterprise to sign on to Spectre, the computer vision fruit sizing tool developed by New Zealand start-up, Hectre.
"Spectre uses computer vision, machine learning and AI technology to detect fruit size, all from the simple click of an iPad or iPhone. Sample sizes captured by Spectre are 100 times those obtained in traditional sizing practices and accuracy levels are extremely high. Results are served up within seconds making Spectre the fastest and most accurate, portable computer vision citrus sizing tool available on the market," explains Hectre CEO, Matty Blomfield.
Using the Spectre for Citrus app, size distribution graphs are served up within seconds of taking the photo using just an iPad or iPhone.
Launched in 2020 to the US and NZ apple industries, Spectre is now being used in some of the world’s largest apple growing regions, such as Washington. Clients include apple giants, Washington Fruit and Produce Co. and Yakima Fruit and Cold Storage.
Multiple use case application is one of the key factors leading to the hype around Spectre.
“Spectre can be used at grower level in the field for pre-store size assessment, QC integration, and fruit size tracking, right down to lot and block level. For fruit warehouses, huge value is gained with Spectre at receiving, with quick and easy capture of size data reducing the incidence and negative impacts of incorrect storage, packing line resets, and wasted labour and fruit,” says Bloomfield.
The simplicity of Spectre belies the complex nature of the technology, and has been a cornerstone of Hectre’s overall offering which includes Hectre Orchard Management Software. For Spectre, users only have to stand and raise an iPad or iPhone over a fruit bin, take a photo, and the job is done. Within seconds, fruit size distribution graphs detailing average fruit size and sample size, are served up to the same device. No expensive equipment is required nor reliance on inaccurate human sample selection.
“When we got the call asking if Spectre could size citrus, as well as apples, we jumped at the chance to extend Spectre’s capabilities,” adds Blomfield.
The Hectre R&D team, led by Luke Butters, travelled to Gisborne, New Zealand’s most prolific citrus growing region and collected high volumes of image data for oranges, lemons, mandarins and tangelos.
“In less than two months, orange detection optimisation, ground truthing and grader comparisons for accuracy were complete, and Spectre for Citrus was born, cementing a valued relationship between Hectre and First Fresh NZ,” says Butters.
Thousands of images and the talents of the Hectre R&D team were used to teach the computer about the intricacies of citrus via computer vision and machine learning methodologies, leading to high accuracy rates for Spectre for Citrus.
Spectre for Citrus is now being released to the US and Australian citrus markets and the company is excited to see what the year will bring, especially as citrus gains an even stronger foothold in the global healthy food market.
Hot on the heels of Spectre for Citrus, has been the exclusive release of Spectre for Colour, computer vision colour detection with the same simple and speedy user experience. Defect detection is what industry is now requesting of Hectre and the company has confirmed the project has high placement on their product development roadmap.
For more information: