Recently, convolutional neural networks (CNNs), which exhibit excellent performance in the field of computer vision, have been in the spotlight. However, as the networks become wider for higher accuracy, the number of parameters and the computational costs increase exponentially. Therefore, it is challenging to use deep learning networks in embedded environments with limited resources, computational performance, and power. Moreover, CNNs consume a great deal of time for inference. To solve this problem, we propose a practical method for filter pruning to provide an optimal network architecture for target capacity and inference acceleration. After revealing the correlation between the inference time and the FLOPs, we proposed a method to generate a network with the desired inference time. Various object detection datasets were used to evaluate the performance of the proposed filter pruning method. The inference time of the pruned network was measured and analyzed using the NVIDIA Jetson Xavier NX platform. As a result of pruning the number of parameters and FLOPs of the YOLOv5 network in the PASCAL VOC dataset by 30%, 40%, and 50%, the mAP decreased by 0.6%, 2.3%, and 2.9%, respectively, while the inference time was improved by 14.3%, 26.4%, and 34.5%, respectively.

The objective of this study is to monitor the water content of soil quickly and accurately using a UAV. Because UAVs have higher spatial and temporal resolution than satellites, they are currently becoming more useful in remote sensing areas. We developed a water content estimation equation using the color of the soil and suggested a calibration method for field application. Since the resolution of the images taken by the UAV is different according to the altitude, the water content estimation formula is developed by using the images taken at each altitude. In order to calibrate the color difference according to lighting conditions, the calibration method using field data were proposed. The results of the study showed an altitude-specific estimation equation using RGB values of the UAV image through linear regression. The appropriate number of field data needed for calibration for site application of the estimation equation was found between 4 and 10. On-site application results of the proposed calibration method showed RMSE accuracy of 1.8 to 2.9%. Thus, the water content estimation and calibration method proposed in this study can be used in effectively monitoring the water content of soil using UAVs.

No abstract

No abstract

For appropriate managing fields and crops, it is essential to understand soil properties. There are drawbacks to the conventional methods currently used for collecting a large amount of data from agricultural lands. Convolutional neural network is a deep learning algorithm that specializes in image classification, and developing soil property prediction techniques using this algorithm will be extremely beneficial to soil management. We present the convolution neural network models for estimating water content and dry density using soil surface images. Soil surface images were taken with a conventional digital camera. The range of water content and dry density were determined considering general upland soil conditions. Each image was divided into segmented images and used for model training and validation. The developed model confirmed that the model can learn soil features through appropriate image argumentation of few of original soil surface images. Additionally, it was possible to predict the soil water content in a situation where various soil dry density conditions were considered.

No abstract

Recent developments in drone technology have led to the widespread use of unmanned aerial vehicles (UAVs). In particular, UAVs are often used in reconnaissance to detect objects such as missing persons in large areas. However, traditional systems use only one UAV to search for missing persons in a large area. In addition, object detection is performed after flight or manually because detection requires high computing power. In this paper, a reconnaissance drone system using multiple UAVs is proposed. The proposed multi-UAV reconnaissance system performs real-time object detection on each UAV. The realtime object detection results from each UAV are received by the ground control system (GCS) to stitch the images. To enable real-time object detection in individual UAVs, the filter pruning method is applied to the YOLOv5 model, and the model uses 40% fewer parameters than the existing baseline model. The lightweight YOLOv5 model achieves approximately 11.73 FPS on the Jetson Xaiver NX using a mission computer. Moreover, the proposed image stitching method enables image stitching by effectively matching features using additional information generated by UAVs. The UAV flight tests show that the proposed reconnaissance system can monitor and detect objects in real time over large areas.INDEX TERMS Image stitching, network pruning, real-time object detection, swarm flight system.

Generally, soil moisture and salinity in reclaimed land are monitored using soil dielectric sensors such as time domain reflectometry, frequency domain reflectometry, and capacitance. The soil dielectric sensor measures apparent dielectric permittivity. However, apparent dielectric permittivity is affected by soil moisture, salinity, and texture. In this study, performance evaluation and calibration of a dielectric sensor (5TE; METER Group, Inc., Pullman, WA, USA) for monitoring soil salinity were performed. Laboratory calibration tests were completed, incorporating various levels of dry density, water content, and salinity. The soil salinity was determined by the electrical conductivity (EC) 1:5 method. The volumetric water content as measured by the sensor was affected by dry density and water content. Generally, it linearly increased as dry density and water content increased. However, when dry density or water content was high, the measured value of the sensor increased nonlinearly. The bulk EC measured by sensor had no specific correlation with EC 1:5. The EC 1:5 measurement had a linear relationship with the gradient of θ and θ s. Therefore, it can be estimated with a simple linear equation using θ from the soil test and θ s from the capacitance sensor. The R 2 value of the EC 1:5 estimation equation was 0.98. The proposed equation requires θ from the gravimetric sample and θ s from the sensor. Therefore, in the case of monitoring salinity using a sensor, it is recommended to measure the water content with a tensiometer.