computer vision, grain counting, sorghum


An estimation of on-farm yield before harvest is important to help farmers make decisions about additional input use, time to harvest, and options for end uses of the harvestable product. However, obtaining a rapid assessment of on-farm yield can be challenging, especially for a sorghum (Sorghum bicolor L.) crop due to the complexity of counting the total number of grains in a panicle at field-scale. One alternative to reduce labor is to develop a rapid assessment method employing computer vision algorithms. Computer vision has already been utilized to account for the number of grains within a panicle, yet it has only been tested under controlled conditions. The objective of this study was to estimate the number of grains in a sorghum panicle using imagery data captured from a smartphone device at field-scale. During the pre-harvest season, sorghum panicles of several commercial hybrids were photographed in the field. Later, the plants corresponding to those panicles were harvested to determine the final number of grains, to develop a benchmarking dataset. Using Python language and the OpenCV library, each image was filtered, blurred, and contours were applied to estimate the number of grains in each sorghum panicle. The absolute mean difference obtained using the algorithm output for the observed and the estimated number of grains was 570 (root mean square percentage error = 53%).


Rights Statement

In Copyright - Educational Use Permitted.

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.