Harvest and post-harvest techniques for soybean grains and seeds using interactive machine learning

By Ricardo Aparecido Santos, professor at the Technological College of the State of Goiás

08.09.2022 | 16:47 (UTC -3)

In seed production, an extremely important item is the quality of the product, as it depends on the processing levels and operational performance in an organization. When it comes to grains, their quality is linked to the conditions in which the product breaks down in full conservation and the conditions of the agricultural market.

In this sense, we can differentiate the issues of managing the harvest and post-harvest of grains and seeds from farming to shipping. In seed production, for example, it differs from grains, as in cultivation the monitoring levels in relation to seeds is a more intensified activity, which requires 5 to 6 samples to be taken, requiring constant inspection by the Ministry of Agriculture, Livestock and Supply (MAPA). In inspections, it takes into account agronomic attributes in field conditions, incidence of invasive plants and stink bugs, physical characteristics linked to water content and mechanical changes in the tegument and seed quality.

Harvest point and monitoring

When harvesting seeds, the ideal moisture point for soybeans is between 16 and 18%. When it comes to grains, harvesting is done at a time when humidity is below 15%. To do this, it is necessary to adjust the harvester, checking the moisture conditions of seeds and grains, taking into account the inspection of the percentage of seeds with a seed coat that has suffered physical damage, and the seeds must not exceed 8%. An extremely important feature for seeds is monitoring in trucks and harvesters, although the same does not occur in grain cultivation, which leads to conserving the genetic attributes of cultivars.

One factor that needs to be taken into consideration is from transportation to receipt, as the same influences the quality of seeds more than grains. Therefore, it is important to analyze the occurrences of heating and proliferation of fungi that harm the initial stages of the seeds, which happens in periods that exceed 12 hours.

The processes linked to the pre-cleaning of grains and seeds occur mainly at the time of harvest, with the use of harvesters that are widely regulated and operated, and there may be a percentage of impurities above the minimum levels that are appropriate for conservation and negotiation.

Receiving strategies and operational flow speed

The processes linked directly to sampling aim to measure the moisture of a given batch of grains. To do this, it is necessary to train the workforce correctly to collect the grain mass. Therefore, the samples are classified, in accordance with the Normative Instruction of the Ministry of Agriculture, Livestock and Supply (MAPA) no. 11/2007.

In grain production, for example, there is official and commercial classification, with the commercial classification being established in the form of a contract between the buyer and seller. The buyer defines the quality model that will be implemented by the farmer, and if the agreed limit level is exceeded, it will be subject to a deduction in the value of the product. As for seeds, moisture assessments, presence of a unique mixture, percentage of mechanical changes must be carried out in the company's extensively sanitized laboratory, taking into account the quality of the product.

Decision-making is an extremely important element in seed production, as the water content has the same influence on the drying conditions of seed lots. In decisions, for example, they are managed together with the reception and prior verification teams, being a condition for analyzing issues of uniformity of the physiological quality of the batch, percentage of mechanical change, percentage of water content and percentage of greenish seeds.

Drying and Storage Technologies

In non-natural storage drying of grains and seeds, the objective is to remove the amount of water, allowing better conditions for the conservation of chemical, physical and physiological properties for the storage of plant products.

To dry grains and seeds, they must have humidity conditions greater than 13%. When storing soybeans, it is considered safe when it has water contents of 11 and 12% at the end of the drying process.

When drying soybean grains and seeds, one of the most used methods is continuous drying, as the acceleration of drying is very different between grains and seeds. In grains, much higher drying air temperatures are used; in the case of seeds, the temperature cannot be higher than 38ºC at the end of the drying process.

As a result, soybeans have been one of the main crops in the agricultural sector globally. Its seeds are rich mainly in lipids, amino acids, vitamins and minerals, being a crop that presents a diversification of products of extreme importance for food security in the international market. For the crop to be successful, it is necessary to use seeds with high germination and quality conditions.

As for seed quality, it is directly related to environmental conditions and the insertion of post-harvest procedures, with the use of machines and modified drying. Such factors, when talking directly about seeds, take into account damage and appearance, influenced by pathogen attack, changes in humidity, greenish seeds and seed breakage, which can change their performance. As a result, many efforts have been made to improve the efficiency and quality of seeds, especially with regard to locating lots, monitoring, and inserting chemical methods.

Therefore, the use of machine learning has been supported directly by computer vision. This approach is characterized by the use of machine learning oriented to interactive machine users. These Interactive Machine Learning actions aim to work with the use of algorithms that interact in both computational and human conditions, promoting the use of skills and knowledge to provide support and precision to the models.

When talking directly about Interactive Machine Learning, it aims to solve problems with small data interactions or complex information when conventional machine learning is inefficient. Such machine learning technologies with computer vision made it possible to analyze the quality of products of plant origin, especially with regard to seeds, being a support tool for visual inspection in the area of ​​agricultural product technology.

Finally, the use of interactive machine learning to classify the quality of soybean seeds is of paramount importance, as it has the possibility of solving several problems linked to plant production and its various models, which aims to be able to simulate possible events that will emerge in real time. Therefore, it is necessary to qualify professionals in the field of seeds to solve the adversities that occur in the world of agribusiness.

By Ricardo Aparecido Santos, professor at the Technological College of the State of Goiás

Mosaic Biosciences March 2024