Classification of Crop Residue Cover in High-Resolution RGB Images Using Machine Learning

HighlightsA machine learning framework estimated residue cover in RGB images taken at three resolutions from 88 locations.The best results primarily used texture features, the RFE-SVM feature selection method, and the SVM classifier.Accounting for shadows and plants plus modifying and optimizing the texture features may improve performance.An automated system developed using machine learning is a viable strategy to estimate residue cover from RGB images obtained with handheld or UAV platforms.Abstract. Maintaining plant residue on the soil surface contributes to sustainable cultivation of arable land. Applying machine learning methods to RGB images of residue could overcome the subjectivity of manual methods. The objectives of this study were to use supervised machine learning while identifying the best feature selection method, the best classifier, and the most effective image feature types for classifying residue levels in RGB imagery. Imagery was collected from 88 locations in 40 row-crop fields in five Missouri counties between early May and late June in 2018 and 2019 using a tripod-mounted camera (0.014 cm pixel-1 ground sampling distance, GSD) and an unmanned aerial vehicle (UAV, 0.05 and 0.14 GSD). At each field location, 50 contiguous 0.3 × 0.2 m region of interest (ROI) images were extracted from the imagery, resulting in a dataset of 4,400 ROI images at each GSD. Residue percentages for ground truth were estimated using a bullseye grid method (n = 100 points) based on the 0.014 GSD images. Representative color, texture, and shape features were extracted and evaluated using four feature selection methods and two classifiers. Recursive feature elimination using support vector machine (RFE-SVM) was the best feature selection method, and the SVM classifier performed best for classifying the amount of residue as a three-class problem. The best features for this application were associated with texture, with local binary pattern (LBP) features being the most prevalent for all three GSDs. Shape features were irrelevant. The three residue classes were correctly identified with 88%, 84%, and 81% 10-fold cross-validation scores for the 2018 training data and 81%, 69%, and 65% accuracy for the 2019 testing data in decreasing resolution order. Converting image-wise data (0.014 GSD) to location residue estimates using a Bayesian model showed good agreement with the location-based ground truth (r2 = 0.90). This initial assessment documents the use of RGB images to match other methods of estimating residue, with potential to replace or be used as a quality control for line-transect assessments. Keywords: Feature selection, Soil erosion, Support vector machine, Texture features, Unmanned aerial vehicle.

Dry Bean (Phaseolus vulgaris L.) Crop Water Production Functions and Yield Response Factors in an Arid to Semi-Arid Climate

HighlightsDeficit irrigation negatively affected dry bean yield and yield components.Excess irrigation increased crop ETc but not dry bean yield.Soil moisture fluctuation was greater in the top 0.3 m of the soil profile compared to deeper depths.Crop water production function had a slope of 18.9 kg ha-1 mm-1 and threshold crop evapotranspiration of 171 mm.Dry bean crop was found to be sensitive to water stress (yield response factor Ky = 1.94).Abstract. Under changing climate conditions and declining water resources, understanding crop response to water stress is critical for effective irrigation management. The objectives of this study were to quantify dry bean (Phaseolus vulgaris L., cv. Othello) soil moisture dynamics, crop evapotranspiration (ETc), and yield response factor and to develop dry bean irrigation and crop water production functions (IWPF and CWPF). Five irrigation treatments, i.e., full irrigation (FIT), 75% FIT, 50% FIT, 25% FIT, and 125% FIT, were evaluated using a randomized complete block design (RCBD) with three replications for three years (2017, 2018, and 2019) in the arid to semi-arid intermountain region of Powell, Wyoming. The results showed a significant influence of irrigation on dry bean soil moisture dynamics and ETc. The dry bean crop showed a greater soil moisture fluctuation in the top 0.3 m of the soil profile compared to 0.6 m and at 0.9 m. ETc ranged from 187 to 438 mm, from 190 to 409 mm, and from 217 to 398 mm in the 2017, 2018, and 2019 growing seasons, respectively. A positive two-segment relationship was observed between dry bean seed yield and cumulative irrigation water applied. The average cumulative seasonal irrigation of 310 mm resulted in maximum seed yield. For all three years, the seed yield increased linearly with ETc. Combining the data from the three years resulted in a CWPF with a slope of 18.9 kg ha-1 mm-1 and an offset of 171 mm of ETc (i.e., the ETc required for crop establishment before any seed yield is produced, or threshold ETc). Moreover, the dry bean crop was found to be sensitive to water stress (Ky = 1.94). These results indicated that under the typical semi-arid to arid climate conditions of the intermountain region of Wyoming, deficit irrigation of dry bean may not be a viable strategy because the yield loss outweighs water-saving benefits. Keywords: Dry bean, Crop evapotranspiration, Crop production function, Irrigation water production function.

Characterization of the Composition, Structure, and Mechanical Properties of Endocarp Biomass

HighlightsEndocarps have higher lignin content, cellular and bulk density, and hardness than typical biomass feedstocks.The impacts of lignin content, bulk density, and mechanical properties on energy consumption are discussed.Endocarps can be a potential feedstock for a biorefinery coproducing biofuel and bioproducts.Abstract. Lignin is an abundant biopolymer and a promising source of feedstock for high-value chemicals and materials. This study aims to characterize the lignin-rich endocarp biomass and identify features of this unique feedstock that are relevant to feedstock preprocessing and logistics. The chemical composition and cellular structure of walnut and peach endocarps were characterized using HPLC and scanning electron microscopy (SEM) imaging. Mechanical properties of the endocarps were investigated using nanoindentation. Mechanical tests revealed hardness values of up to 0.48 and 0.40 GPa for walnut and peach endocarps, respectively. With screen sizes of 1 and 2 mm, the specific energy consumption was 9.21 and 1.86 MJ kg-1 for walnut and 12.6 and 2.72 MJ kg-1 for peach, respectively, as determined using a knife mill. Milling energy consumption was correlated to screen size, lignin content, bulk density, and mechanical properties. This study provides critical information on feedstock supply logistics necessary to implement a novel feedstock in biorefineries and evaluate the economic feasibility for coproduction of biofuels and lignin-derived products. Keywords: Biomass feedstock, Lignin, Mechanical properties, Nanoindentation, Size reduction.

Evaluation of Design Characteristics for Animal Mortality Composting Systems

HighlightsDesign characteristics for animal mortality compost cover materials were tested.Compressive stress was applied to simulate the effects of the mortalities on cover materials.The highest permeability was measured for sawdust at 25% moisture content.A linear relationship was found between the volumetric flow rate and the power required to aerate the piles.Abstract. Composting is an aerobic process that relies on natural aeration to maintain proper oxygen levels. Air-filled porosity, mechanical strength, and permeability are among the essential parameters used to optimize the process. This study’s objective was to measure the physical parameters and airflow characteristics of three commonly used cover materials at four moisture levels, which could be used in designing actively aerated swine mortality composting systems. A laboratory-scale experiment was conducted to measure pressure drops across the cover materials as a function of the airflow rate and the material’s moisture content. Compressive stress was applied for 48 h to simulate the impact of swine mortalities on the cover materials. The power required to aerate each material was determined as a function of volumetric flow rate and moisture content. As expected, air-filled porosity and permeability decreased with increasing bulk density and moisture content. The highest average permeability values were measured at 25% moisture content and ranged from 66 × 10-4 to 70 × 10-4 mm2, from 161 × 10-4 to 209 × 10-4 mm2, and from 481 × 10-4 to 586 × 10-4 mm2 for woodchips, ground cornstalks, and sawdust, respectively. For the range of airflow rates tested in this study (0.0025 to 0.0050 m3 s-1 m-2), a linear relationship (R2 = 0.975) was found between the volumetric flow rate (m3 s-1) and the power required to aerate the compost pile (W per 100 kg of swine mortality). Keywords: Airflow, Darcy’s law, Livestock, Modeling, Permeability, Pressure drop.

Tobacco Leaf Grading Based on Deep Convolutional Neural Networks and Machine Vision

HighlightsA classification model for the front and back sides of tobacco leaves was developed for application in industry.A tobacco leaf grading method that combines a CNN with double-branch integration was proposed.The A-ResNet network was proposed and compared with other classic CNN networks.The grading accuracy of eight different grades was 91.30% and the testing time was 82.180 ms, showing a relatively high classification accuracy and efficiency.Abstract. Flue-cured tobacco leaf grading is a key step in the production and processing of Chinese-style cigarette raw materials, directly affecting cigarette blend and quality stability. At present, manual grading of tobacco leaves is dominant in China, resulting in unsatisfactory grading quality and consuming considerable material and financial resources. In this study, for fast, accurate, and non-destructive tobacco leaf grading, 2,791 flue-cured tobacco leaves of eight different grades in south Anhui Province, China, were chosen as the study sample, and a tobacco leaf grading method that combines convolutional neural networks and double-branch integration was proposed. First, a classification model for the front and back sides of tobacco leaves was trained by transfer learning. Second, two processing methods (equal-scaled resizing and cropping) were used to obtain global images and local patches from the front sides of tobacco leaves. A global image-based tobacco leaf grading model was then developed using the proposed A-ResNet-65 network, and a local patch-based tobacco leaf grading model was developed using the ResNet-34 network. These two networks were compared with classic deep learning networks, such as VGGNet, GoogLeNet-V3, and ResNet. Finally, the grading results of the two grading models were integrated to realize tobacco leaf grading. The tobacco leaf classification accuracy of the final model, for eight different grades, was 91.30%, and grading of a single tobacco leaf required 82.180 ms. The proposed method achieved a relatively high grading accuracy and efficiency. It provides a method for industrial implementation of the tobacco leaf grading and offers a new approach for the quality grading of other agricultural products. Keywords: Convolutional neural network, Deep learning, Image classification, Transfer learning, Tobacco leaf grading

Optimization and Testing of Seed Box Parameters of Potato Planters Based on Seed Potato Group Motion Simulation

HighlightsThe structural parameters of the seed box of a cup-belt potato planter were optimized.The arching phenomenon of seed potato in the reseeding box was reduced.The qualified rate of cup-belt potato planters in field operation was improved.Abstract. To solve the problem of missed sowing caused by the arching phenomenon of seed potato in the reseeding box of a cup-belt potato planter, this study used the discrete element method to analyze the seed potato movement in the seed box. The analysis revealed unreasonable structural parameters of the seed box. Therefore, the discrete element method was used to optimize the structural parameters of the seed box. First, the structural parameters of the seed box were analyzed by single-factor analysis, and then quadratic rotation orthogonal combination simulation tests of three factors and five levels were performed. Two evaluation indexes were established to evaluate the flow performance of the seed box: the amount of seed taken minus the amount of seed replenishment (CD), and the coefficient of variation of the average seed-population velocity in the reseeding box (CV). The best factor combination was found through optimization. With a 42° inclination angle of the bottom surface of the seed box (a), 145 mm reserved height of the seed isolation baffle (h), and 100° inclination angle of the seed isolation baffle (ß), the CD was 2.622 grains, and the CV was 22.887%. With these parameters, the mobility of seed potato in the reseeding box was good, and there was no arching in the reseeding box. To verify the accuracy of the optimized results, a cup-belt potato seeding test bench was built to conduct verification tests. The bench test verified the rationality of the model and the optimized parameters. Field tests showed that the arching phenomenon in the reseeding box was solved by optimizing the structural parameters of the seed box. The purpose of improving the sowing performance of the cup-belt potato planter was achieved. Keywords: Agricultural machinery, Discrete element method, Parameter optimization, Potato planter, Seed box.

Effect of Post-Drying Tempering of Rice on Minimizing Kernel Fissuring and Maximizing Moisture Removal

HighlightsMore moisture can be removed in a single drying pass without severely fissuring kernels when samples are tempered than when immediately cooled without tempering.Tempering rice kernels immediately after drying can reduce the percentage of fissured kernels by up to half of that when kernels are immediately cooled without tempering.Abstract. Improper rice drying results in kernel fissuring, leading to head rice yield reduction due to breakage during milling. The objective of this study was to determine the percentage points (pp) of moisture content (MC) reduction that can be achieved in a single drying pass without significantly fissuring kernels. Long-grain rough rice of cultivars CL XL745 and Diamond at initial MCs of 18%, 17%, 16%, 15%, and 14% were dried using air at 45°C/20% relative humidity (RH), 50°C/15% RH, 55°C/12% RH, 60°C/10% RH, and 65°C/8% RH to MCs of 17%, 16%, 15%, 14%, 13%, or 12% with and without post-drying tempering. All temperature/RH combinations resulted in a humidity ratio of 0.012 kg water kg-1 dry air. Tempering was conducted at the drying air temperature for 4 h. The resulting samples achieved between 1 and 7 pp of MC reduction in a single drying pass. The pp of MC reduction that can be attained in a single drying pass without causing significant fissuring varied across the cultivars tested. Generally, ~2 pp of MC reduction was achieved in a single drying pass for CL XL745 and ~4 pp for Diamond without causing adverse fissuring when samples were not tempered after drying. However, with tempering, ~3.5 pp of MC reduction was achieved in a single drying pass for CL XL745 and ~5.5 pp for Diamond without causing significant fissuring. However, these amounts varied depending on the drying air conditions and initial MC. For both cultivars, tempering immediately after drying reduced the fissured kernel percentage by up to half of that when the kernels were not tempered. These findings quantify the importance of rice tempering and provide information on how much moisture can be safely removed in a single drying pass. Such findings may be applied to different dryer types to reduce fissuring due to drying, thereby minimizing head rice yield reductions. Keywords: Drying, Glass transition, Rice quality, Single-pass drying, X-ray imaging.

Design and Experimental Study of a Spiral Auxiliary Feeding Device for Lodged Corn on a Combine Harvester

HighlightsThis study designed an auxiliary feeding device for lodged corn on a combine harvester.The mechanical characteristics of lodged corn stalks were studied as the basis of the design.The working parameters were optimized in an orthogonal experiment.Abstract. Lodging causes kernel and ear loss in mechanical harvesting of corn and threatens grain security. To address this issue, the mechanical bending properties and lifting forces of lodged corn stalks were explored in this study. The ears of lodged corn could be lifted to the normal harvesting height by applying an upward force to the stalks. Based on this conclusion, an auxiliary feeding device for lodged corn in a combine harvester was designed. An orthogonal optimization experiment was conducted to explore the effects of working parameters on the device performance. The results demonstrated that the sequence of the factors affecting kernel loss with the auxiliary feeding device was: harvesting direction, forward speed of the harvester, and rotating speed of the spiral stalk lifter. The optimal combination of harvester forward speed and rotating speed of the spiral stalk lifter was 0.5 m s-1 and 300 rpm, and the ratio of the peripheral speed to the speed of motion was 9.42. The optimal harvesting direction was opposite to the lodging direction of the stalks. This study provides a theoretical basis and design reference for low-loss harvesting of lodged corn. Keywords: Auxiliary feeding, Combine harvester, Corn, Harvest loss, Lodging, Orthogonal experiment.