查看更多>>摘要：? 2022 Elsevier B.V.Rice can provide humans with basic energy and nutrients, and its quality has attracted great attention. Under different storage conditions, the rice quality is different. In the work, a depth discrimination model, ultra-lightweight dynamic attention network (ULDAN), is proposed to realize rice quality identification combined with electronic nose (e-nose) technology. Firstly, e-nose is used to collect rice gas information under different humidity conditions. Secondly, Ultra-lightweight dynamic convolution block (ULDC) is proposed to extract e-nose data features, which changes the traditional convolution kernel calculation method to enhance the representational ability of features. Thirdly, the convolution classification layer (CCL) is introduced to replace the average pooling layer and fully connected layer to compress the parameter amount and improve the classification performance. In conclusion, ULDAN obtains good classification results under different humidity, of which 75% RH is the best. In addition, these results can provide the technology for quality control.
查看更多>>摘要：? 2022 Elsevier B.V.Leaf chlorophyll content (LCC) is one of important indicators for photosynthesis evaluation and crop nutrition diagnosis. For establishing an accuracy LCC diagnosis model, this study focused on reduce influences of multi-source interferences which resulted from external noises and internal multi-compositions. A systematic strategy with pipelines methods of hybrid preprocessing and wavelengths optimization was proposed to overcome such complicated interferences. Maize datasets of canopy reflectance and LCC measurements were collected under six fertilization treatments and three replications in three growth stages (V6, V9, and V12) in 2019 summer maize. Hybrid preprocessing was proposed by cascading preprocessing methods of discrete wavelet transformation (DWT) smoothing, multiplicative scatter correction (MSC), second derivative (D2). Optimization of sensitive wavelength was conducted to reduce the influences of multi-components overlapping. Methods of competitive adaptive reweighted sampling (CARS), iterative retaining informative variable (IRIV), and variable iterative space shrinkage space (VISSA) combined with partial least square (PLS) were used to explore and validate spectral responses traits of multi-components overlapping. The results of VISSA-PLS showed superior potentials for diagnose maize LCC with accuracy of RP2 of 0.76 and RMSEP of 2.45. The modeling results indicated that the systematic solution of DWT-MSC-D2-VISSA-PLS could improve the performance of LCC diagnosis. It provided application potentials for accurate diagnosis of maize growth status and lay foundations for topdressing recommendations.
查看更多>>摘要：? 2022Tackling the grand challenges of 21st century agriculture (Ag) will require a fundamental shift in the way we envision the role of artificial intelligence (AI) technologies, and in the way we build agricultural AI systems. This shift is needed especially for complex, high-value agricultural ecosystems such as those in the Western U.S., where 300+ crops are grown. Farmers and policy makers in this region face variable profitability, major crop loss and poor crop quality owing to several challenges, including increased labor costs and shortages of skilled workers, weather and management uncertainties, and water scarcity. While AI is expected to be a significant tool for addressing these challenges, AI capabilities must be expanded and will need to account for human input and human behavior – calling for a strong AI-Ag coalition that also creates new opportunities to achieve sustained innovation. Accomplishing this goal goes well beyond the scope of any specific research project or disciplinary silo and requires a more holistic transdisciplinary effort in research, development, and training. To respond to this need, we initiated the AgAID Institute, a multi-institution, transdisciplinary National AI Research Institute that will build new public-private partnerships involving a diverse range of stakeholders in both agriculture and AI. The institute focuses its efforts on providing AI solutions to specialty crop agriculture where the challenges pertaining to water availability, climate variability and extreme weather, and labor shortages, are all significantly pronounced. Our approach to all AgAID Institute activities is being guided by three cross-cutting principles: (i) adoption as a first principle in AI design; (ii) adaptability to changing environments and scales, and (iii) amplification of human skills and machine efficiency. The AgAID Institute is conducting a range of activities including: using agricultural AI applications as testbeds for developing innovative AI technologies and workflows; laying the technological foundations for climate-smart agriculture; serving as a nexus for culturally inclusive collaborative and transdisciplinary learning and knowledge co-production; preparing the next generation workforce for careers at the intersection of Ag and AI technology; and facilitating technology adoption and transfer.
查看更多>>摘要：? 2022 Elsevier B.V.Plantain tree is the most popular crop grown all over the world and banana (Musa spp.) is the most marketable fruit. It is the leading food in many countries, especially in developing countries. Plant diseases are significant aspects that result in a serious reduction in the quantity and quality of fruit crops. Plantain tree cultivation is affected by various diseases such as Black Sigatoka/Yellow sigatoka, Panama, Bunchy top, Moko, chlorosis, etc. Rapid and novel approaches for the apt discovery of diseases help farmers in developing better decisions and efficient control measures. Convolutional Neural Networks (CNN) and Recurrent Neural Network (RNN) have been proved their efficiency in several fields and it has recently moved in the field of crop disease classification and detection. The objective of this research work is to create a Deep Learning Model for the disease classification and its early prediction to support farmers in plantain tree cultivation. A new sequential image classification model is proposed to detect the diseases by combining RNN and CNN, which is named as Gated-Recurrent Convolutional Neural Network (G-RecConNN). The input to the proposed model is the sequences of plant images. The experiments are carried out in real-time datasets collected from the state named Tamil Nadu situated in the Southern part of India. This method aims at numerous advantages such as reduced pre-processing of the data, easy online performance evaluation and advancements with less real data, etc. The experimental results inspired the utilization of the G-RecConNN model with farmer support systems that will process continuous banana tree images as part or whole for the early detection of banana tree diseases.
查看更多>>摘要：? 2022 Elsevier B.V.Digitalization and automation of assessments in field trials are established practice for farming product development. The use of image-based methods has provided good results in different applications. Although these models can leverage some problems, they still perform poorly under real field conditions using mobile devices on complex applications. Among these applications, insect counting and detection is necessary for integrated pest management strategies in order to apply specific treatments at early infection stages to reduce economic losses and minimize chemical usage. Currently the counting task for the assessment of the degree of infestation is done manually by the farmer. Current state of the art object counting methods do not provide accurate counting in crowded images with overlapped or touching objects which is the case for insect counting images. This makes necessary to define novel approaches for insect counting. In this work, we propose a novel solution based on deep learning density map estimation to tackle insects counting in wild conditions. To this end, a fully convolutional regression network has been designed to accurately estimate a probabilistic density map for the counting regression problem. The estimated density map is then used for counting whiteflies in eggplant leaves. The proposed method was compared with a baseline based on candidate object selection and classification approach. The results for alive adult whitefly counting by means of density map estimation provided R2 = 0.97 for the counted insects in the main leaf of the image, that outperforms by far the baseline algorithm (R2 = 0.85) based on image processing methods for feature extraction and candidate selection and deep learning-based classifier. This solution was embedded to be used in mobile devices, and it has been gone for exhaustive validation tests, with diverse illumination conditions and background variability, over leaves taken at different heights, with different perspectives and even unfocused images, for the analyzed pest under real conditions.
查看更多>>摘要：? 2022Crop row following is especially challenging in narrow row cereal crops, such as wheat. Separation between plants within a row disappears at an early growth stage, and canopy closure between rows, when leaves from different rows start to occlude each other, occurs three to four months after the crop emerges. Canopy closure makes it challenging to identify separate rows through computer vision as clear lanes become obscured. Cereal crops are grass species and so their leaves have a predictable shape and orientation. We introduce an image processing pipeline which exploits grass shape to identify and track rows. The key observation exploited is that leaf orientations tend to be vertical along rows and horizontal between rows due to the location of the stems within the rows. Adaptive mean-shift clustering on Hough line segments is then used to obtain lane centroids, and followed by a nearest neighbor data association creating lane line candidates in 2D space. Lane parameters are fit with linear regression and a Kalman filter is used for tracking lanes between frames. The method is achieves sub-50 mm accuracy which is sufficient for placing a typical agri-robot's wheels between real-world, early-growth wheat crop rows to drive between them, as long as the crop is seeded in a wider spacing such as 180 mm row spacing for an 80 mm wheel width robot.
查看更多>>摘要：? 2022During the seedling stage, real-time monitoring and detection of seed germination are important for testing the quality of seeds, crop field management, and yield estimation. However, owing to the low efficiency of traditional manual seedling rate monitoring, survey methods have been gradually replaced by unmanned aerial vehicles (UAVs) and real-time peanut video counting models. In this study, we propose an efficient and fast real-time peanut video counting model (combining the improved YOLOV5s, DeepSort, and OpenCV programs) to accurately distinguish peanut seedlings from weeds, and to count peanut seedlings based on videos. The improved YOLOV5s combines a vision transformer with CSNet to replace the original CSNet backbone. The field experiment results show that the real-time peanut video counting model count capabilities is close to those of humans with an accuracy of 98.08%; however, the seedling calculation model takes only one-fifth of the time required for human detection. Therefore, the video-based model outperforms the image-based target detection algorithm, and was more suitable for application in practical germination rate investigation in peanut production.
查看更多>>摘要：? 2022 The Author(s)The generalisation performance of a convolutional neural network (CNN) is influenced by the quantity, quality, and variety of the training images. Training images must be annotated, and this is time consuming and expensive. The goal of our work was to reduce the number of annotated images needed to train a CNN while maintaining its performance. We hypothesised that the performance of a CNN can be improved faster by ensuring that the set of training images contains a large fraction of hard-to-classify images. The objective of our study was to test this hypothesis with an active learning method that can automatically select the hard-to-classify images. We developed an active learning method for Mask Region-based CNN (Mask R-CNN) and named this method MaskAL. MaskAL involved the iterative training of Mask R-CNN, after which the trained model was used to select a set of unlabelled images about which the model was most uncertain. The selected images were then annotated and used to retrain Mask R-CNN, and this was repeated for a number of sampling iterations. In our study, MaskAL was compared to a random sampling method on a broccoli dataset with five visually similar classes. MaskAL performed significantly better than the random sampling. In addition, MaskAL had the same performance after sampling 900 images as the random sampling had after 2300 images. Compared to a Mask R-CNN model that was trained on the entire training set (14,000 images), MaskAL achieved 93.9% of that model's performance with 17.9% of its training data. The random sampling achieved 81.9% of that model's performance with 16.4% of its training data. We conclude that by using MaskAL, the annotation effort can be reduced for training Mask R-CNN on a broccoli dataset with visually similar classes. Our software is available on https://github.com/pieterblok/maskal.
查看更多>>摘要：? 2022 Elsevier B.V.Due to the increase of global demand of agriculture and agro-industrial products, agriculture has undergone some changes in the stages of procurement, planning, planting, maintenance, harvesting, transport and distribution. The use of formal mathematical model has been employed to evaluate and implement improvement policies and strategies. In this paper, we present a Mixed Integer Nonlinear Programming Problem model to solve the planning of agricultural production systems in the stages of sowing, crop maintenance and harvesting. The objective function seeks to minimize the cost of resources and wastes generated in operations. The minimization is inspired by lean manufacturing wastes but newly applied in agricultural production. The proposed approach is validated using data for the planning of bananas farmers in the Caribbean region of Colombia. While a deterministic solution is obtained, sensitivity analyses are carried out to evaluate different scenarios for production based on surface response methodology. Model offers a reduction of cost near 59% of cost relationed with wastes of production, moreover, we can find an alternative for scheduling operations in agricultural production based in the most important resources and tasks required along of production in agriculture.
查看更多>>摘要：? 2022 Elsevier B.V.Accurate recognition of cherry tomatoes is a key issue for the automatic picking system in plant factories, which helps to improve picking efficiency and reduce production costs. By using the depth information and considering the prior adjacent constraint between the fruit and the stem of cherry tomatoes, this paper proposes an improved Mask R-CNN for visual recognition of cherry tomatoes. Firstly, the input layer of the network is modified to achieve dual-mode data fusion of RGB and depth images. Secondly, by constructing the corresponding region generation network to indicate the integral constraint between the fruit and the stem, false recognition of branches is reduced. Thirdly, a multi-class prediction subnetwork is used to decouple the pixel-level category predictions of fruit and stem. Meanwhile, multi-task loss balance and adaptive feature pooling are adopted to overcome the limitation caused by the size difference between fruit and stem. The experimental results show that the improved Mask R-CNN achieved an accuracy of 93.76% for fruit recognition, which is 11.53% and 15.5% higher than that of the standard Mask R-CNN and YOLACT, and it achieves an accuracy of 89.34% for stem recognition, which is 13.91% and 19.7% higher than that of the standard Mask R-CNN and YOLACT, respectively. Besides, the recall rate of the proposed method for stem recognition is 94.47%, which is 11.53% and 8.3% higher than that of YOLACT and Mask R-CNN, respectively. In addition, the proposed method takes only 0.04 s to process a single image, providing an efficient approach for automatically picking cherry tomatoes in plant factories.
NSTL
Elsevier