Search

Internet of Things (IoT) has been implemented in most advance technologies as part of the emerging 4th industrial revolution, and recently, blockchain technology is being welcomed. These rapid growing technologies giving a big challenge to the students. As such, learning these new technologies should be made easy and simple. However, current education does not specifically offer a curriculum to empower skills and knowledge in IoT and blockchain technology. The aim of this research is to develop learning kit that can provide suitable training to understand concept of IoT and blockchain technology. The kit consists of three parts namely “brain”, “muscle” and “cloud”. Raspberry Pi is used as “brain” of operation that will be interact with the Caiser Cloud platform. A testbox operates as “muscle” to provide simple data input. The input from testbox will be sent, stored and displayed on a platform created using Xojo software. The results show that the learning kit is successfully interact with Caiser Cloud platform and can be used as a training tools for education purpose.

The current practice in determining oil palm fresh fruit bunches (FFB) ripeness is by its colour which could be inaccurate. This study investigates the classification of oil palm FFB ripeness using Raman spectroscopy. A feature extraction model is developed based on the different organic compositions that contribute to the ripeness classification. Samples are collected according to the Malaysian Palm Oil Board (MPOB) standards which are unripe, underripe, ripe, overripe, and rotten. Different characteristics of the Raman shift were detected which represent the material composition for each sample. The Raman intensity of the oil palm fruit increases from unripe to ripe before decreasing to rotten due to the carotenoid content in the fruit. In conclusion, Raman spectroscopy is a suitable technique to observe the changes in the composition of oil palm fruit classified by its ripeness.

The use of attributes in person re-identification and video surveillance applications has grabbed attentions of many researchers in recent times. Attributes are suitable tools for mid-level representation of a part or a region in an image as it is more similar to human perception as compared to the quantitative nature of the normal visual features description of those parts. Hence, in this paper, the preliminary experimental results to evaluate the robustness of attribute detectors against pose and light variations in contrast to the use of local appearance features is discussed. Results attained proven that the attribute-based detectors are capable to overcome the negative impact of pose and light variation towards person re-identification activities. In addition, the degree of importance of different attributes in re-identification is evaluated and compared with other previous works in this field.

Nutrient management in oil palm plantation is considered as one of the prominent issues especially for smallholder farmer. The nutrient contained in the tress has always been neglected and untreated and these may cause the trees to suffer from nutrient deficiencies. Therefore, in leveraging the oil yield at the maximum, a telemonitoring system is developed to assess and monitor the lack of nutrients for respective trees. This is done using image processing technique and artificial intelligence in detecting the nutritional deficiencies by analyzing the leaf. The categorization focused by classifying into four major types either as magnesium deficiencies, potassium deficiencies, nitrogen deficiencies or healthy that is based on the oil palm’s leaf surface. This is achieved by extracting the features namely number of red pixels, entropy and correlations. Further, two classifiers specifically support vector machine and artificial neural network is used for classification purpose along with performance measure using accuracy(ACC), Mean Square Error (MSE), Mean Absolute Error (MAE), Sensitivity (SN), Specificity (SP), Positive Predictive Value (PPV), Negative Predictive Value (NPV) based on ten-fold cross-validation. Results attained showed that the best classifier is SVM using RBF kernel (SVM-RBF) that is capable to accurately recognize the nutrient deficiencies with 100% accuracy.

Most 3D scanners are heavy, bulky and costly. These are the major factors that make them irrelevant to be attached to a drone for autonomous navigation. With modern technologies, it is possible to design a simple 3D scanner for autonomous navigation. The objective of this study is to design a cost effective 3D indoor mapping system using a 2D light detection and ranging (LiDAR) sensor for a drone. This simple 3D scanner is realised using a LiDAR sensor together with two servo motors to create the azimuth and elevation axes. An Arduino Uno is used as the interface between the scanner and computer for the real-time communication via serial port. In addition, an open source Point-Cloud Tool software is used to test and view the 3D scanner data. To study the accuracy and efficiency of the system, the LiDAR sensor data from the scanner is obtained in real-time in point-cloud form. The experimental results proved that the proposed system can perform the 2D and 3D scans with tolerable performance.

The objective of this paper is to optimise the proposed control strategy for an active anti-roll bar system using non-dominated sorting genetic algorithm (NSGA-II) tuning method. By using an active anti-roll control strategy, the controller can adapt to current road conditions and manoeuvres unlike a passive anti-roll bar. The optimisation solution offers a rather noticeable improvement results compared to the manually-tuned method. From the application point of view, both tuning process can be used. However, using optimisation method gives a multiple choice of solutions and provides the optimal parameters compared to manual tuning method.

This work presents a qualitative measurement of anterior segment photographed images (ASPIs) to identify between normal eyes and eyes with pterygium and pinguecula through Otsu multi-thresholding approach without contrast enhancement. In addition, we also propose a mobile screening framework of ASPIs through smartphones. ASPIs were directly sent to the cloud storage once an ASPI was captured using a smartphone camera, and then each image was processed through a digital image processing approach in a processing platform. Three important steps, namely, pre-processing, image segmentation and qualitative assessment, are involved in the processing platform of the mobile screening framework. The ASPIs are pre-processed to minimise or eliminate any unwanted areas within the image. Then, these ASPIs are segmented through multi-thresholding Otsu approach with clustering number n = 3. Segmentation result shows that the accuracy of the proposed method is 87.5%, which is comparable with the previously established work that has applied three-step differencing (3SD) method. However, the proposed approach has better computational time which is six times faster than the 3SD method. These results demonstrate a remarkable effort to produce a simple but straightforward digital image processing approach to be implemented in cloud computing for future studies.

This paper presents a proposed supervised classification technique namely partial histogram Bayes (PHBayes) learning algorithm. Conventional classifier based on Gaussian function has limitation when dealing with different probability distribution functions and requires large memory for large number of instance. Alternatively, histogram based classifiers are flexible for different probability density function. The aims of PHBayes are to handle large number of instances in datasets with lesser memory requirement, and fast in training and testing phases. The PHBayes depends on portion of the observed histogram that is similar to the probability density function. PHBayes was analyzed using synthetic and real data. Several factors affecting classification accuracy were considered. The PHBayes was compared with other established classifiers and demonstrated higher accurate classification, lesser memory even when dealing with large number of instance, and faster in training and testing phases.

Crop management is one of the key aspects in precision agriculture. On-site crop management includes a scheduled prediction and prescription of fertilization and pesticide application on specific areas of the field. A prediction of the tree health is much needed in order to decide a suitable prescription for the plant. An autonomous vehicle, equipped with at least one LiDAR (light detection and ranging) sensor, could be used not only for detecting and mapping its surrounding, but to also help evaluate tree health for early distress detection. Currently, farmers have to rely on eyesight to identify trees or plants in distress. Big-scale plantations depend on costly scheduled airborne monitoring routines, which also relies on human vision by scouring through hours of aerial video footage. Both techniques have similar weaknesses in terms of the time it takes to detect a tree in distress and the accuracy of the detection using human vision. The objective of this research is to propose a technique in evaluating tree health using a simple LiDAR sensor that is commonly used in autonomous vehicles. In order to achieve this objective, an evaluation of the different intensity characteristics of tree leaves versus fruits was carried out, both in the lab and in the field. This study has chosen oil palm trees as its subject, as the problem of health monitoring in oil palm plantations is evident. A LiDAR with a 905 nm near-infrared (NIR) laser is used to scan both individual healthy leaves in the lab and different oil palm trees in the field. Since a LiDAR sensor is normally used for ranging, a systematic process was proposed to capture the reflected intensity value of the laser beam that was transmitted to the object. This whole system can be realized using a LiDAR sensor, servo motors, and an Arduino board. Processing software was used to test and store the captured information from the sensor. Later, MATLAB was used to plot the intensity map of the leaves and oil palm tree, classify the range intensity into histograms, and calculate the leaf area index (LAI) for the oil palm trees. From the experimental results, it is found that the reflectance intensity of the leaves shows consistent range values between 155-160 magnitude both in the lab and in the field. From there, three different trees with different number of leaves were scanned and evaluated based on their LAI values. The health of the tree is then predicted, where a healthy tree is estimated to have a higher LAI value. The resulting LAI value is found to correlate with the evaluation using eyesight. This proves that although using a single-wavelength NIR laser beam provided by the LiDAR sensor, as compared to multiple wavelengths of a spectrometer, the difference between the oil palm leaves, fruits and the background noise could be determined. In the future, where multi-wavelength laser LiDAR sensors could be possible, more materials could be characterized. In conclusion, detection, mapping, and materials characterization could be done by an autonomous vehicle utilizing a LiDAR sensor, where tree health could be predicted for crop care management.