Farms.com Home   Ag Industry News

How computer vision and deep learning benefits yield management

How computer vision and deep learning benefits yield management

Bloomfield Robotics new custom stereo camera features active lighting and stereo lenses.

By Ainsley Andres, Farms.com

Attendees of the 2021 Farms.com virtual Precision Agriculture Conference & Ag Technology Showcase got an exclusive look at Bloomfield Robotics’ custom stereo camera, designed to collect high quality images in any agricultural environment.

Ben Utter, Business Development Director at Bloomfield Robotics, presented the technology as an innovative way for growers to collect and measure key plant attributes for high-throughput phenotyping. It is a combination of ground-based image capturing, deep learning-based processing, and computer vision artificial intelligence.

Unlike fixed cameras, Bloomfield’s camera features active lighting and stereo lenses. “The active lighting allows you to collect data at any time of the day or night, while the stereo lens’ provides depth perception in particular,” explained Utter.

The camera can be attached to any vehicle moving through a farm and collects images at human scale to extract measurements such as yield, maturity, or disease incidence.

“Once the image is captured, the idea is to digitally turn every single aspect of the plant into usable and actionable information,” Utter said.

Amongst the company’s focus on vineyards, its technology can also be applied to different kinds of crops. Bloomfield has tried its technology in apple orchards, cherries, and sorghum, and is currently working in Peru with a blueberry grower.

“We’re very interested on being able to translate the technology to other crops,” stated Utter. “The crucial thing to think about is the value—what kinds of problems are they experiencing in that crop? And is our technology suited to help try and solve and tackle those challenges?”

In addition, Bloomfield was selected by NASA to receive funding based on their technical merit and commercial potential. A big part of that grant is pursing Bloomfield’s edge processing capabilities.

Utter explained that Bloomfield’s mission is to maintain being inexpensive and easy to use. This means sticking to the RGB visual spectrum, which already provides valuable information, but it is also experimenting with hyperspectral imaging, as part of their recent grant from NASA.

Watch his presentation in its entirety below:




Trending Video

Grass-Cast: Reducing Risk In Times Of Drought

Video: Grass-Cast: Reducing Risk In Times Of Drought

Grass-Cast is an online tool that helps ranchers and other land managers anticipate drought earlier and respond with more flexibility and foresight. Earlier drought detection and informed action can reduce risk and stabilize economic returns. Grass-Cast was developed by a team of USDA and university partners with outreach led by the USDA Northern Plains Climate Hub.
 

Comments


Your email address will not be published