カテゴリー: 会誌106号_Vol.10

特別小特集　Artificial Intelligence of Things (AIoT) for Smart Farming 2
　
Oil Palm Fresh Fruit Bunch Detection and Ripeness Classification Using YOLOv5
モバイル端末で自動診断スマート農業技術の実用例

Mohamed Yasser Mohamed Ahmed Mansour, Katrina D. Dambul, and Kan Yeep Choo

In oil palm estates, smart farming technology such as visual-based mobile applications can improve the harvesting process by assisting in the oil palm fresh fruit bunch (FFB) ripeness classification process and allow the monitoring of the harvesting process remotely. In this paper, oil palm FFB detection and ripeness classification is developed using a YOLOv5s model. Augmentation and label smoothing techniques were applied to improve the performance of the model. The results after applying label smoothing techniques showed that the mean average precision of the YOLOv5s model is 87.85%(0.5:0.95) with a precision of 96.19% and a recall of 95.19%. The results after image augmentation (Mosaic and Cut-Out) showed that the mean average precision of the YOLOv5s model is 86.67%(0.5:0.95) with a precision of 95.78% and a recall of 98.00%. This model can potentially be implemented in a mobile device to be used in a real time harvesting process.

特別小特集　Artificial Intelligence of Things (AIoT) for Smart Farming 3
　
Mm-wave and THz Scanning for Non-invasive Farm Product Safety
ミリ波/THz 波を利用した異物混入のリアルタイム検出技術

Nguyen NGOC MAI-KHANH, Shintaro TAKADA, Keizo INAGAKI, Tran NGOC LE, Tran THI MY HANH, Hinano SUGIMOTO, Akio HIGO, Hitoshi ABATA, Makoto IKEDA, Bich-Yen NGUYEN, and Tetsuya KAWANISHI

Our project is to develop a portable and cost-effective scanner for real-time detection of foreign contaminants in agricultural staples and/or animal feed with a unique and novel mm-wave/THz sensing capability. This paper presents both qualitative and quantitative detection methods for the harmful substance melamine in a wide frequency range, from mm-wave to the THz range. Melamine has been illegally forcibly added to meals, including powdered milk or pet feed, in order to raise the protein level, one of the most crucial quality indicators. A method for quantitative detection of melamine is presented in this paper. Validation in both mm-wave and THz ranges is performed using experiments. The measurement results show that the estimation of melamine content in mixtures can be performed in several mm-wave and THz frequency bands, including 220-330GHz, 2THz, 2.26THz, and 4THz. Especially the peak around 4THz of melamine was found and to our knowledge this is the first study and THz measurement. To further evaluate the properties and response of melamine and its mixtures in the infrared spectrum and other chromatography, the quantitative determination of melamine was performed by both Fourier-transform infrared spectroscopy (FTIR) method and ultra-performance liquid chromatography (HPLC). These frequency bands of melamine property are useful for implementing a portable detection system with an integrated antenna array or THz pulse transceiver since the size of the components and the sampling resolution essentially depend on the operating frequency band or wavelength.

特別小特集　Artificial Intelligence of Things (AIoT) for Smart Farming 4
　
Wireless Sensor Network Optimization in Agriculture:Brief Review and Case Study Using 2.4GHz Transceivers
電力消費を考慮して広域な農場をカバーするセンサ配置の求め方

Wei Kitt WONG, Saaveethya SIVAKUMAR, Filbert H. JUWONO, and Ing Ming CHEW

Internet of things (IoT) has been widely applied in various sectors, including in agricultural sector for crop monitoring. To cover a wide area, the deployment of a wireless sensor network (WSN) is required to transmit data to a single access point. The sensors, however, have a few constraints, including energy consumption and service coverage (range). As a result, it is critical to optimize the configuration of the sensors to obtain the desired performance. With this in mind, the objective of this paper is threefold. First, the state-of-the-art of the WSN-IoT in agricultural sector is presented. subsequently, some of the optimization techniques corresponding to the WSN are discussed along with some areas that need to be considered in the optimization technique. Lastly, to solidy the the entire discussion, a case study of optimizing WSN topology parameters in order to minimize the power consumption of the NRF24L01＋transceivers is presented. In particular, Adaptive Differential Evolution (ADE) algorithm with normalization-binarization approach is proposed to encode the adjacency matrix for optimizing the power consumption.

特別小特集　Artificial Intelligence of Things (AIoT) for Smart Farming
　
Editorial Preface

Editorial Team Leader Yee-Loo Foo

特別小特集　Artificial Intelligence of Things (AIoT) for Smart Farming 1
　
Image-based Classification of Leaf Diseases Using Convolutional Neural Networks
葉病害を早期検知畳込みニューラルネットワークによる葉病分類

Muhammad Umair, Wooi-Haw Tan, and Yee-Loo Foo

Smart farming is on the uprising demand all over the world. Artificial Intelligence (AI) is considered as one of the latest tools to be utilized for smart farming. However, the practical implementation of AI for smart farming is often a challenge. One of the challenges is to optimize the algorithms for accurate classification of plant diseases. In this study, we have proposed a Convolutional Neural Network (CNN) for the classification of leaf diseases. The framework of the proposed CNN is designed using the Depthwise Separable Convolution (DWS) technique that consists of two stages, i. e., depthwise and pointwise feature extractions. We have compared the model with the classical convolutional approach. Results show that the proposed model outperformed the conventional CNN model with a precision of 0.932, recall 0.992, F1 score of 0.961 and a test accuracy of 95.25% whereas the conventional model achieved precision 0.941, recall 0.961, F1 score 0.951 and 93.76% of test accuracy.