Afdhal Mahamooth | Undergraduate | Informatics Institute of Technology (IIT)

Computer vision and pattern recognition are computer science fields that focus on teaching computers to understand and grasp information by training machines to extract meaningful information and identify patterns and regularities in data through processing and analysis. The use of such technologies is in dire need in the agricultural industry, mainly in a developing country like Sri Lanka, which has a varied climate and rain pattern for a small country and a scarcity of effective commercial fertilizers to withstand the harsh conditions that nature has to offer. Therein lies the necessity to automate agriculture. Farming with the help of computer vision and pattern recognition provides the ability to achieve the benefits of being cost-effective and efficient.

Typically, crop growth monitoring relies on human judgment, which is not always reliable regardless of how experienced the said individual may be, for there are other factors that contribute to it, such as environmental factors, etc. Crop monitoring is an important stage in computer vision and pattern recognition in agriculture. Information is gathered at different stages to understand the growth environment and to optimize the growth with certain changes. Using technology to monitor crops is highly efficient in comparison to non-automated, archaic techniques. Real-time monitoring with computer vision and pattern recognition would identify slight changes in the product caused by factors such as malnutrition, etc., much sooner than the human eye. The output quality of vital crops such as rice and wheat influence the stability and food security of a country. Therefore, the ability to consistently monitor plant development and responsiveness to nutritional requirements is critical.

MATLAB was used to extract leaf features from different leaf locations and assess the dynamic properties of rice leaves to diagnose nitrogen levels by Yongbin Sun, Haibin Duan, and Ning Xian. To measure the blade variation process, newly established parameters such as the yellowing area (EA), degree of yellowing (ED), and form (area and perimeter), as well as color attributes (green, standardized red index, etc.), are employed. Wheat heading date is one of the most essential criteria for wheat crops. Investigation of an autonomous computer vision surveillance system for the wheat heading period in order to precisely record the heading date was carried out by Yanjun Zhu, Zhiguo Cao, Hao Lu, Yanan Li, and YangXiao. The testing findings demonstrated that the approach was clearly superior to the old method, with the absolute error of the test dataset being 1.14 days [2]. Though the method was sturdy, it had tiny errors, and only the heading period could be monitored. To monitor the heading and flowering stages of wheat in the growth stage, Pouria Sadeghi-Tehran, Kasra Sabermanesh, Nicolas Virlet, and Malcolm J. Hawkesford developed a new method [3].The new method albeit its limitations have better accuracy and the ability to be used in challenging environmental conditions. Though it had its own limitations, the new method was the most effective.

We could conclude by saying that computer vision and pattern recognition will be a vastly useful technology in the world to further the efficiency of agriculture and increase the quality of the produce, thereby leading to economic stability.