Search

In soil, plant residues have low contrast making them difficult to detect using X‐ray computed tomography. In this work, we tested a convolutional neural network (U‐Net) for its ability to improve the identification of crop residues in soil samples assembled from aggregates of different size fractions (small, large, water‐stable aggregates, and average aggregate composition). Soil CT images were obtained using a 244 μm resolution. About 2500 soil images were annotated to train the neural network, of which only 631 images were selected for the training data set. Intersection over Union (IOU) was used as a measure of success of segmentation by neural network, which takes values from 0 to 1. In the validation data set, IOU of background was 0.93, IOU of solid phase was 0.95, IOU of pore space was 0.77, and IOU of plant residues was 0.40. However, IOU of plant residues in the total data set increased to 0.7. Soil structure influences the quality of multiphase segmentation of soil CT images. The poorest segmentation of plant residues was in the soil samples composed of average aggregate size composition. The quality of pore space segmentation increased with increasing porosity of the soil sample. The model tends to generalize the large areas occupied by plant residues and overlooks the smaller ones. The low values of the IOU metric for plant residues in the training data set can also be related to insufficient quality of annotation of the original images. U‐Net convolutional neural network segments plant residues in soil.The ability to segment soil into solids, pores, and plant residues by neural networks depends on the soil structure type.In the validation data set, the quality of background segmentation by the Intersection over Union (IOU) measure was 0.93, IOU of the solid phase was 0.95, and IOU of the pore space was 0.77.The quality of segmentation for plant residues by the IOU measure was 0.4 and 0.7 in the validation and total data set, respectively.

The work explores methods and develops aids to improve a pilot flight training level using Full Flight Simulators. It examines their didactic advantages over real helicopters by implementing on-board intelligent decision support systems (IDSS). These simulation programs monitor actions of learner, find the best ways to correct errors and generate appropriate instructions. The training is carried out by reference point methods while the helicopter flight dynamics model simulates a flight and helps an instructor in parallel. Sufficiently simple and accurate model of helicopter flight dynamics was selected and tested to assess the proposed methods and aids. Our analysis introduces a functional structure of IDSS as an adaptive control system with a reference model. We further conducted a study on developing models, methods and means of automatic analysis, forecasting, optimization and correction of actions. Combining single-criteria conditional and vector optimization methods, we found out optimal flight parameters. The research findings revealed that the chosen optimization method requires too many system resources for the optimal solution to be found in a short time. The study thereby comes up with applying ANNs to solve this problem.

Core IdeasSmartphone cameras can be used to reliably measure soil color.Soil color measured with smartphone cameras had less subjectivity and uncertainty.Natural lighting conditions affected the accuracy of the soil color measured with smartphone cameras. Soil color is one of the most useful soil properties for assessing and monitoring soil health. Here we present results of tests of a new soil color app for mobile phones. Various smartphone cameras (SPCs) were tested under sunny and cloudy conditions and compared with visual estimates using Munsell color charts (MCCs). The measured and estimated soil colors were then compared with the “true” colors determined using a spectrophotometer. The results indicated that soil color determinations based on SPC measurements under both sunny and cloudy conditions were as good as those obtained using the MCCs. The accuracy of the SPC measurements was affected by the natural illumination conditions, with higher accuracy in the sun than where clouds were present. Our results also indicated that the SPC measurements completed in the sun provide higher precision (lower variance) than SPC measurements completed under cloudy conditions or estimates based on MCCs. These results suggest that mobile‐device cameras have great potential to allow non–soil scientists, and others lacking access to color charts, to determine soil color.