https://www.iprep.edu.au/round-1-2018Startup Glenrowan Enterprises is leading a collaboration applying Artificial Intelligence (AI) in precision agriculture, using neuromorphic hardware for identification of weeds and pests and their location in paddock. The technology, branded "Sniper - One Shot" is being developed in a collaboration involving Glenrowan Enterprises, Brainchip, Cisco Innovation Centre, Curtin and ECU, Farmers - Ryan Farming Company, Agronomists - Agvivo, machinery OEMs and the Department of Primary Industry and Regional Development (DPIRD). The project utilises BrainChip's patented software and hardware technologies and a low cost Red, Green, Blue visible light camera sensor to classify items identified in the images. As the machinery (combine harvester or boom spray) traverses the paddock, a Global Positioning System is used to determine the location where the weed, crop or pest was identified and a map is developed. The initial proof of concept validation will utilise data from DPIRD's Royalties for Regions funded project into mapping patches of weeds and snail s by monitoring grain contaminants during harvesting. Herbicide resistance, invasive species and pests present a real cost to farmers and regulators who manage them.
project title
Using AI to identify small conical snails, annual ryegrass and three horned bedstraw within crop sample visible images
project summary
Initially a data set of 1000 and then a further 5000 sample images will be provided by DPIRD specifically for the identification of small conical snails, annual ryegrass and three horned bedstraw within crop samples. Imagery data will need to be processed through the Brainchip software package. To accelerate this process, we require the development or modification of an existing API to achieve batch processing on laptop/desktop.
It is expected that the output data will yield further classifications of other crop seeds and contaminants which will require manual QA review by a qualified agronomist / agricultural scientist. Further image data sets will need to be obtained from the Grain Trade Australia and Australian Plant Phenolics Facility (Adelaide) to build out a library of seeds and crop/weeds and using this to determine the key areas to focus on that yield the greatest economic impact to farmers.
A commercial product concept will need to be developed that would involve the high-level system design and specification. The concept product is to be designed to be mounted on the clean grain elevator housing on a CASE 8230 or John Deere S680/780 harvester to allow further development as the BrainChip hardware becomes available.
(20min delay)