Father of genetic science Gregor Mendel spent age boringly mention and measuring pea plant industrial plant trait by script in the 1800s to uncover the basics of genetic inheritance . Today , plant scientist can track the traits , or phenotypes , of hundreds or K of plant life much more quickly , with automated camera systems . Now , Salk researchers have helped bucket along up plant phenotyping even more , with machine - learning algorithmic rule that instruct a estimator organization to analyze three - dimensional form of the branches and leaves of a plant . The sketch , publish in Plant Physiologyon October 7 , 2019 , may help scientists better quantify how plants respond to clime modification , genetic variation or other factors .
“ What we ’ve done is acquire a suite of tools that helps turn to some common phenotyping challenges , ” says Saket Navlakha , an associate professor in Salk ’s Integrative Biology Laboratory and Pioneer Fund Developmental Chair .
A flora ’s environment helps prescribe its structure , which is related to its health . scientist trying to understand plant growth , direct more resilient plant or boost crop production often want to quantify elaborated characteristics of a plant ’s leaves and shoot . To do this phenotyping in a high - throughput way , many researchers use camera system that take image of each flora from various angles and assemble a three - dimensional model . However , some mensuration are surd to take with these sew - together images .
Recently , some have turn to a new method acting , called 3D optical maser scanning , to entrance the structure of plant architectures . Researchers shine a optical maser at each plant to “ paint ” its airfoil with the shaft . The resulting data — called a 3D point cloud — portrays the hunky-dory detail of the plant ’s aerofoil . But quantitatively psychoanalyse the point clouds can be challenging since the technology is so young and the datasets so enceinte .
“ The resolution and accuracy of this information is much higher , ” says Navlakha . “ But the method acting that have been developed for analyzing leave and branch in 2D images do n’t work as well for these 3D point befog . ”
Navlakha , along with UC San Diego graduate student Illia Ziamtsov , used a 3D laser scanner to skim 54 Lycopersicon esculentum and baccy plants grown in a mixed bag of conditions . Then , they inputted the leave 3D point clouds into automobile - acquire algorithmic program that let them teach the programme how to phenotype the flora . The technique involved the researchers first indicating manually where leaves and shoots on the flora were . Then , the software began to automatically recognize these features .
A Salk technician 3-D scan a plant life .
“ It ’s like teaching things to a baby , ” says Navlakha . “ You give them examples of what a foliage appear like and what a branch looks like , and eventually they can discover a plant they ’ve never seen before and pick out the parting and outgrowth . ”
The researchers focused on teaching the program to make three phenotype measurements that scientists often use — separating stems from leaf , count leaves and their sizing , and outlining the branching traffic pattern of a plant . They found they were successful : for example , the method had a 97.8 per centum accuracy at identifying radical and leaves .
“ This sort of object detection has been used in self - drive cars and for identifying construction and furniture items , ” says Ziamtsov . “ But applying it to works is entirely fresh . ”
Navlakha and Ziamtsov need to continue fine - tune up the plan of attack ; differentiating two tightlipped - together leaves can still be challenging , for example . And the current reading of the package may not work on all types of plants . They hope to generalize the software to work on plants from vine to trees , and also to analyze roots .
“ There are a draw of challenges in agriculture right now to try and increase crop product and sequester atomic number 6 better , ” say Navlakha . “ We go for our tool can help biologists address some of these broader challenge . ”
Navlakha and Ziamtsov will release their software as open - source for other research worker to use . They hope the software will rush up industrial plant research , since it makes eminent - throughput phenotyping faster and easier .
“ Doing this variety of analysis by hand is very laborious , ” says Ziamtsov . “ Our tool does it quickly and fairly accurately . ”
Source : Salk Institute for Biological Studies
