Search Results (3)

Herein, we present the novel method targeted for determination of plant nutritional state with the use of computer vision and Neural Networks. The method is based on multispectral imaging performed by an exclusively designed Agroscanner and a dedicated analytical system for further data analysis with Neural Networks. An Agroscanner is a low-cost mobile construction intended for multispectral measurements at macro-scale, operating at four wavelengths: 470, 550, 640 and 850 nm. Together with developed software and implementation of a Neural Network it was possible to design a unique approach to process acquired plant images and assess information about plant physiological state. The novelty of the developed technology is focused on the multispectral, macro-scale analysis of individual plant leaves, rather than entire fields. Such an approach makes the method highly sensitive and precise. The method presented herein determines the basic physiological deficiencies of crops with around 80% efficiency. Full article

Land use conversion is generally accompanied by large changes in soil organic carbon (SOC). SOC influences soil erodibility through its broad control on aggregate stability, soil structure and infiltration capacity. However, soil erodibility is also influenced by soil properties, clay mineralogy and other human activities. This study aimed to evaluate soil organic carbon as proxy of soil erosion risk in the Nyumba ya Mungu (NYM) catchment in Northern Tanzania. Soil organic carbon (SOC) was measured by an AgroCares scanner from which the soil organic matter (SOM) was derived using the conversional van Bemmelen factor of 1.72. A regression analysis performed between the measured loss on ignition (LOI) values and SOM from the AgroScanner showed a strong positive correlation in all land use classes (LOI_FL R² = 0.85, r = 0.93, p < 0.0001; LOI_CL R² = 0.86, r = 0.93, p = 0.0001; LOI_GL R² = 0.68, r = 0.83, p = 0.003; LOI_BS R² = 0.88, r = 0.94, p = 0.0001; LOI_BL R² = 0.83, r = 0.91, p = 0.0002). This indicates that SOC from the soil scanner provided a good representation of the actual SOM present in soils. The study also revealed significant differences in the soil aggregate stability (WSA) and SOM stock between the different land use types in the Upper Pangani Basin. The WSA decreases approximately in the following order: grassland > forest land > bare land > cultivated > bush land. Land use change can thus potentially increase the susceptibility of soil to erosion risk when SOC is reduced. Since WSA was directly related to SOM, the study indicates that, where formal measurements are limited, this simple and inexpensive aggregate stability test can be used by farmers to monitor changes in their soils after management changes and to tentatively assess SOC and soil health. Full article

Adequate soil information to adapt fertilizer plans and support farmers’ yield ambitions is either hard to obtain or expensive, as it often requires soil sampling and analyses in a lab. AgroCares has developed two services, i.e., the Scanner and the Lab-in-a-box, that place the knowledge of soil analysts and agronomists in the hands of the farmer in a quick, easy and affordable way. The obtained spectral image of the soil provided by the scanner is compared to data in the Global Soil Database; using machine learning regression models, the content of the soil sample is predicted based on its spectrum. The results are returned to the farmer as a soil status report. The Global Soil Database is developed country by country and starts by determining the number and location of the samples required to cover the full spectral range of the specific country using data such as soil type, land use, fertilizer and crop residue management, satellite crop development images, climate and elevation. These samples are then collected following protocols and shipped to the Golden Standard Laboratory in the Netherlands where they are analyzed using regulated, traditional wet chemistry techniques and scanned with the sensors of the Lab-in-a-Box (Mid-Infrared and XRF) and the Scanner (Near-Infrared). The reference values obtained in the GSL and the spectra for each sample obtained from the Scanner and the Lab-in-a-Box form the ground truthing data set required for the machine learning algorithms. Once all the soil data have been extracted from the spectral image, they are sent to the fertilizer module, where the different nutrients are allocated to soil fertility categories. These categories are used to establish the quantities in kg/ha of nutrients needed to reach the desired level of soil fertility. Using local nutrient crop uptake tables, the total nutrient requirements are calculated and converted into fertilizer recommendations that consider factors like nutrient loss after application and available fertilizer. The user then receives a full soil management report that includes the soil analysis results in classes of N, P, K, pH and organic matter with the Scanner, and in values of all macro- and micro- nutrients with the Lab-in-a-box. Full article