Computer vision-based phenotyping for improvement of plant productivity: a machine learning perspective

Abstract

Employing computer vision to extract useful information from images and videos is becoming a key technique for identifying phenotypic changes in plants. Here, we review the emerging aspects of computer vision for automated plant phenotyping. Recent advances in image analysis empowered by machine learning-based techniques, including convolutional neural network-based modeling, have expanded their application to assist high-throughput plant phenotyping. Combinatorial use of multiple sensors to acquire various spectra has allowed us to noninvasively obtain a series of datasets, including those related to the development and physiological responses of plants throughout their life. Automated phenotyping platforms accelerate the elucidation of gene functions associated with traits in model plants under controlled conditions. Remote sensing techniques with image collection platforms, such as unmanned vehicles and tractors, are also emerging for large-scale field phenotyping for crop breeding and precision agriculture. Computer vision-based phenotyping will play significant roles in both the nowcasting and forecasting of plant traits through modeling of genotype/phenotype relationships.

A Weakly Supervised Deep Learning Framework for Sorghum Head Detection and Counting

“…Machine learning and image processing have proved their utility in diverse fields. Especially in the field of plant phenotyping [2][3][4][5][6][7], these tools have laid a strong foundation in detecting multiple crop diseases [8] as well as making sense of disease severity without the need for any additional human supervision [8], crop/weed discrimination [9][10][11][12], canopy/individual extraction [13,14], fruit counting/flowering [15][16][17], and head/ear/panicle counting [18][19][20]. Our hypothesis is that machine learning and image processing along with unmanned aerial vehicles (UAV) based photogrammetry is a reliable alternative to the laborintensive sorghum head survey in the field [21][22][23].…”

Genebank Phenomics: A Strategic Approach to Enhance Value and Utilization of Crop Germplasm

“…The 2D and 3D structural models can be reconstructed from RGB, multispectral, and thermal imagery to derive important agronomic traits for various crops under different environments such as flowering time of rice [ 58 ] and wheat [ 43 ]; crop biomass of field peas [ 42 ] and wheat [ 41 ]; plant height and biomass of rice [ 59 ] and barley [ 60 ]; seed characteristics of lentils [ 61 ], rice [ 62 ], and field peas [ 63 ]; architectural and physiological properties of apple trees [ 64 ]; height and morphological characteristics of blueberries [ 65 ]; canopy temperature of black poplars [ 66 ]; bunch architecture of grapevines [ 67 ]; and ripeness estimation [ 68 ] and fruit counts [ 69 ] of mangos. Recent advances in computer algorithms and machine learning have significantly improved the throughput of raw data processing and analysis, where the processing pipelines have enabled data capture, analysis, and extraction of multiple patterns and features simultaneously [ 70 ]. Machine learning in sensor- and image-based phenotyping has been applied successfully for germination assessment of tomato seeds [ 71 ], head count [ 72 , 73 ], yield prediction in wheat [ 74 ], and prediction of seed longevity in oilseed rape from chemical compositions [ 75 ].…”

Sorghum Panicle Detection and Counting Using Unmanned Aerial System Images and Deep Learning

“…Image segmentation is commonly the first step to extract information of targets from an image by separating a set of pixels containing the objects of interest ( Mochida et al., 2018 ). The application of image segmentation for plant phenotyping is typically implemented at small scales because the input requires detailed information with accurate labels, which is time-consuming and labor-intensive.…”