Diverse pheromones and pheromone-based traps, as well as images acquired from insects captured by pheromone-based traps, have been studied and developed to monitor the presence and abundance of pests and to protect plants. The purpose of this study is to construct models that detect three species of pest moths in pheromone trap images using deep learning object detection methods and compare their speed and accuracy. Moth images in pheromone traps were collected for training and evaluation of deep learning detectors. Collected images were then subjected to a labeling process that defines the ground truths of target objects for their box locations and classes. Because there were a few negative objects in the dataset, non-target insects were labeled as unknown class and images of non-target insects were added to the dataset. Moreover, data augmentation methods were applied to the training process, and parameters of detectors that were pre-trained with the COCO dataset were used as initial parameter values. Seven detectors—Faster R-CNN ResNet 101, Faster R-CNN ResNet 50, Faster R-CNN Inception v.2, R-FCN ResNet 101, Retinanet ResNet 50, Retinanet Mobile v.2, and SSD Inception v.2 were trained and evaluated. Faster R-CNN ResNet 101 detector exhibited the highest accuracy (mAP as 90.25), and seven different detector types showed different accuracy and speed. Furthermore, when unexpected insects were included in the collected images, a four-class detector with an unknown class (non-target insect) showed lower detection error than a three-class detector.

The black pine bast scale, M. thunbergianae, is a major insect pest of black pine and causes serious environmental and economic losses in forests. Therefore, it is essential to monitor the occurrence and population of M. thunbergianae, and a monitoring method using a pheromone trap is commonly employed. Because the counting of insects performed by humans in these pheromone traps is labor intensive and time consuming, this study proposes automated deep learning counting algorithms using pheromone trap images. The pheromone traps collected in the field were photographed in the laboratory, and the images were used for training, validation, and testing of the detection models. In addition, the image cropping method was applied for the successful detection of small objects in the image, considering the small size of M. thunbergianae in trap images. The detection and counting performance were evaluated and compared for a total of 16 models under eight model conditions and two cropping conditions, and a counting accuracy of 95% or more was shown in most models. This result shows that the artificial intelligence-based pest counting method proposed in this study is suitable for constant and accurate monitoring of insect pests.

No abstract

Crop monitoring is a very important area of precision agriculture and smart farming. Through an accurate monitoring, it is possible to more efficiently manage the irrigation, fertilization, and pest control. In this study, we propose aerial thermal image calibration method and thermal image processing techniques to analyze the water stress level of fruit trees under different irrigation conditions. The calibration was performed using Gaussian process regression, and it was demonstrated as an appropriate regression method because it satisfied all requirements including the residuals’ normality, independence, and homoscedasticity. In addition, an appropriate image processing technique was necessary to selectively extract only the canopy temperature from the aerial thermal images, while excluding irrelevant elements such as the soil and other objects. For the image processing techniques, three methods (Gaussian mixture model, Otsu binarization algorithm, and Otsu binarization algorithm after Gaussian blurring) were employed. The Gaussian mixture model provided the highest accuracy and stable results for the extraction of the canopy temperature. After the aerial thermal images were subjected to calibration and image processing, the degree above nonstressed canopy (DANS) water stress index was calculated for the fruit trees under different water supply conditions. The distribution of the DANS water stress index was similar to the distribution of the canopy temperature and inversely proportional to the amount of supplied water content. Therefore, we expect that the DANS water stress index, calculated using the calibration and image processing techniques proposed in this study, can be a reliable measure for the estimation of the water stress of crops for the application of aerial infrared techniques to remote sensing.

Modern people who value healthy eating habits have shown increasing interest in plum (Prunus mume) fruits, primarily owing to their nutritiousness and proven efficacy. As consumption increases, it becomes important to monitor work to prevent Prunus mume fruits from falling out. Moreover, determining the growth status of Prunus mume is also crucial and is attracting increasing attention. In this study, convolutional neural network (CNN)-based deep learning object detection was developed using RGBD images collected from Prunus mume farms. These RGBD images consider various environments, including the depth information of objects in the outdoor field. A faster region-based convolutional neural network (R-CNN), EfficientDet, Retinanet, and Single Shot Multibox Detector (SSD) were applied for detection, and the performance of all models was estimated by comparing their respective computing speeds and average precisions (APs). The test results show that the EfficientDet model is the most accurate, and SSD MobileNet is the fastest among the four models. In addition, the algorithm was developed to acquire the growth status of P. mume fruits by applying the coordinates and score values of bounding boxes to the depth map. Compared to the diameters of the artificial Prunus mume fruits used as the experimental group, the calculated diameters were very similar to those of the artificial objects. Collectively, the results demonstrate that the CNN-based deep learning Prunus mume detection and growth estimation method can be applied to real farmlands.

No abstract

In this study, a portable and large-area blackbody system was developed following a series of processes including design, computational analysis, fabrication, and experimental analysis and evaluation. The blackbody system was designed to be lightweight (5 kg), and its temperature could exceed the ambient temperature by up to 15 °C under operation. A carbon-fiber-based heat source was used to achieve a uniform temperature distribution. A heat shield fabricated from an insulation material was embedded at the opposite side of the heating element to minimize heat loss. A prototype of the blackbody system was fabricated based on the design and transient coupled electro-thermal simulation results. The operation performance of this system, such as the thermal response, signal transfer function, and noise equivalent temperature difference, was evaluated by employing an infrared imaging system. In addition, emissivity was measured during operation. The results of this study show that the developed portable and large-area blackbody system can be expected to serve as a reliable reference source for the calibration of aerial infrared images for the application of aerial infrared techniques to remote sensing.

Highlights Non-destructive soluble solids content prediction model for oriental melon was developed based on NIR spectrum data. Not only the classical ML or Neural-Network methods, but also the mixture of both techniques have also been tried. Comparing the various pre-processing methods, the MSC-PLS-ANN model showed the best results. MSC-PLS-ANN model demonstrated 6% of improvement in RMSE score over the PLSR model, which is commonly used in commercial products Abstract. Models for predicting the soluble solids concentration (SSC) of oriental melons were developed and evaluated by applying near infrared spectroscopy and an artificial neural network technique. For the evaluation, a total of 300 oriental melons, both ripe and unripe, were mixed together and sampled. To develop an SSC prediction model, the actual SSC values of specimens having the same spectra as those of the visible/near infrared wavelength bands were measured. The measured spectra were preprocessed using eight methods [Multiplicative Scatter Correction (MSC), Standard Normal Variate (SNV), Robust Normal Variate, Savitzky-Golay 1st and 2nd; Min-Max Normalization; Robust Normalization; Standardization], and the SSC prediction model was developed by applying three techniques (Partial Least Squared Regression [PLSR], Artificial Neural Network [ANN], and Convolutional Neural Network [CNN]). Among them, the PLSR technique also applied a Variable Importance in Projection (VIP) method for wavelength selection. Among the PLSR-based SSC prediction models, the SNV-preprocessed PLSR model showed the best SSC prediction performance (RMSEtest, 0.67; R2test, 0.81). Among the ANN-based models, the MSC-preprocessed PLS-ANN model showed the best SSC prediction performance (RMSEtest: 0.63, R2test: 0.83). Among the CNN-based models, the DeepSpectra model was applied, but showed the lowest prediction performance (RMSEtest: 0.79, R2test: 0.74). In conclusion, among the three SSC prediction algorithms tested in this study, the PLS-ANN-based prediction model showed the best SSC prediction performance, which was found to be higher than that of the PLSR-based SSC prediction model applied to the sugar sorters currently used in agricultural products at processing centers. Keywords: Artificial Neural Network, Convolution Neural Network, Korean melon, VIP-PLSR, VIS/NIR spectroscopy.