Recent advancements in computer vision, notably in 6D pose estimation with methods like DeepLabv3 and Dense Fusion, show promising results. However, a substantial research gap exists in addressing the limitations of these methods in occluded scenarios, which in need further research to enhance their effectiveness and adaptability in situations involving hidden or partially obscured objects. This study introduces a novel approach employing 6D object pose estimation and pose correction by integrating DeepLabV3 and modified dense fusion. The study exhibits enhanced accuracy in pose estimation and correction for occluded 3D object model x-ray source. Evaluation metrics, including intersection-over-union and mean average, demonstrate high accuracy percentages for detecting the body (98.91%), handle (96.57%), and aperture (89.54%). The mean IoU for each part of the 3D model portable X-ray source ranges from 65.69% to 76.42%. Pose estimation accuracy, assessed through the Average Distance Difference (ADD) metric, indicates superior performance for static pose estimation closer to the camera. Dynamic pose estimation exhibits higher average ADD metrics in scenes with total occlusion. The robustness metric reveals lower lost tracking counts in scenes without occlusion, emphasizing the algorithm's challenges in fully occluded scenarios.

The state of object detection has quickly progressed due to the rapid development of hardware and deep learning models for the said task. The continuous development of the latter allowed its application in real-time. That is, from two-stage to one-stage detectors, which greatly reduced the processing time; and from anchor-based to anchor-free detectors, which allowed a more efficient and flexible model implementation. The purpose of this study is to explore the various YOLO versions available in Ultralytics to detect large office objects for obstacle avoidance. More specifically, this study observed how the versions, 3u, 5mu, 8 and 6m, from which the first three are anchor-free, perform over the dataset. Based from the results, it was found that there were no significant differences between the detectors, except for the anchor-based v6m, which has the worst scores. The four models, v8n, v8s, v8m, and v5mu, were, however, chosen as the candidate models due to their processing speed, ranging from 66 to 75 FPS on the NVIDIA GeForce RTX 3060 GPU. The said models were also subjected to an optimization process in terms of the confidence and IoU threshold values to find the pair that could provide the best metric scores. Results have provided candidate models for the study’s detection task for obstacle avoidance in real-time.

Printed Circuit Boards (PCBs) are vital in modern manufacturing, acting as essential electronic device components and contributing to various technologies' seamless functioning. Due to their significant impact on device performance, any notable flaws in PCBs can lead to operational issues and potential safety hazards. This emphasizes the need for effective defect detection and quality control measures in this crucial electronic component. This research aims to tackle the challenges associated with identifying defects in PCBs by employing deep learning methods. Defects are categorized into functional and cosmetic types, with sub-datasets covering common issues such as missing holes, mouse bites, open circuits, short circuits, spur issues, and spurious copper. Recognizing the variations in PCB designs and the limited availability of defect samples, the project utilizes the You Only Look Once (YOLO) model, which is widely used in object detection. The objective is to train the model to accurately classify defects within the PCB dataset, integrating TensorRT to improve inference program time. The proposed YOLOv8s-TensorRT model had a mean average precision 50 (mAP50) score of 98.5%, which is only 0.1 % lower than the YOLOv8s. Despite that, the integration of TensorRT reduced inference time by 0.9ms and lowered the model size by about 39.3%, both of which are significant in real-time object detection.

The research investigates effective light conditions for the growth of Rauwolfia serpentina, a flower renowned for its medicinal properties. The study meticulously examines the impact of controlling LED grow lights' intensity during both germination and growth phases. To facilitate this investigation, specialized plant nurseries were simulated using the Proteus software and MikroC programming compiler. These nurseries featured black canvas enclosures, PVC pipes, and a PIC16F877A microcontroller equipped with sensors for precise monitoring. By comparing growth outcomes under natural sunlight with those achieved using LED lights with varying exposure durations, the group assessed the plant's efficiency. Notably, the micro controller also controls water pumps, coolers, and a sunroof motor, enhancing environmental control and automation. The controller also features a display used to display the current conditions and status for easier monitoring for the users. The main sensor for the project is the LDR or light dependent resistor which detects how much light is hitting the plants. The findings contribute significantly to the field of controlled environment agriculture, particularly in the context of pharmaceutical plant cultivation.

Aquaponics systems allow simultaneous growth between vegetables and fish with the use of aquatic products. This process is due to the presence of beneficial microbial communities such as nitrifying bacteria. Numerous water quality problems can be addressed by improving the conversion efficiency of biological filtration systems. Maintaining parameters including increased ammonia and nitrite concentration, organic matter accumulation, and decreased levels of dissolved oxygen are imperative to maximize system productivity. The pilot-scale model of the aquaponics system features two independent systems: the control (with biofilter), and the experimental (without biofilter). The media in the biofilter is as follows: an aeration system (for oxygen supply), and a centralized sensory chamber (to facilitate automatic monitoring of pH, temperature, dissolved oxygen, and turbidity). This study assessed the performance of an aquaponic setup with a biofiltration system in terms of reducing the concentration of dissolved solid particles in the water. The correlations of ammonia, nitrate, and nitrate concentrations with turbidity were explored. The turbidity was monitored using a turbidity sensor. To test ammonia, nitrite, and nitrate, solutions were dropped into water samples, whose developed color was compared to a color chart. The system must operate in conditions that cater to the collective growth and development of the fish, plants, and nitrifying bacteria. The plant crop utilized in this study is mint (Mentha spicata), while the fish used was the Red Tilapia (Oreochroomis aureus x Oreochromis mossambicus). Ammonia levels were found to be substantially connected with nitrate levels, and both parameters were shown to be inversely correlated with turbidity. Turbidity in water may be caused by active microorganisms or algae that may metabolize ammonia and nitrate.

The advent of the fourth industrial revolution (4IR) offers promising improvements in operational efficiency and profitability for various industries, and can be a key component to leapfrog the many sectors of manufacturing in the Philippines. This will necessitate the reskilling and upskilling of Filipino automation engineers and instrumentation technicians to commission and maintain smart technologies for production facilities. The further expansion of the country's manufacturing capacities presents an opportunity to locally develop products and solutions for process automation. This includes sensor technologies and applications that can boost the Filipino manufacturers' capabilities through locally-sourced automation components. Various sensor technologies and applications can be explored for further improvement through research & development. Through technology foresight, this study looks into the potential of Filipino technology firms to develop sensors that are locally designed and assembled, addressing the needs of growing industries. The use of scenario-building approach allows for the identification and ranking of the key predictable drivers based on the insights of industry professionals. Opportunities and risks are evaluated based on future possible scenarios.

Chlorophyll content is an imperative indicator of lettuce (Lactuca Sativa) health status. Through computational intelligence, this paper bestowed a noninvasive, accurate, cost-effective ensemble of machine learning algorithms for chlorophyll-b concentration prediction. A total of 107 images of loose-leaf lettuce var. Altima from an aquaponic farm situated in Rizal province in the Philippines was utilized. By employing CIELab color space, the leaf canopies were segmented and extracted with 18-feature predictors. The regression tree ranked and selected 10 selected significant leaf features (spectral: R, G, S, a*, b*, Cr; morphological: canopy area; textural: contrast, correlation, and homogeneity). A fitness function that optimized the recurrent neural network architecture was constructed using GPTIPSv2 which is a symbolic multigene regression (SMGR) tool. This convergence function was the main element in developing a genetic algorithm (GA)-optimized recurrent neural network model considered as the PIGMENTnet. It provides the optimal quantity of neurons in each of the three hidden layers in neural network architecture. A 75-100-10 conglomeration of neurons in each layer was recommended. The RMSE (0.1486), R2 (0.9998), and MAE (0.0751) results of PIGMENTnet surpassed the unoptimized RNN. Based on these findings, it implies that the developed PIGMENTnet is an effective Chl-b concentration predictor as it provided highly accurate and sensitive results than the sole RNN model.

Invasive genetic and chemical-based laboratory techniques are very limited for in situ and in vivo applications especially in classifying leaf species in the wild. Out of 41 recorded Salvia species, 25 are endemic to the Central Asian region. In this study, a non-destructive model for profiling and identifying Salvia species (SalviaNet) was developed by employing computer vision allied with feature-based machine learning. The image set is composed of 25 Salvia species collected over Uzbekistan and other territories (Locus classicus) and photo-scanned to capture the totality of the leaf surface. CIELab thresholding was employed to fully segment the leaf pixels. Classification tree (CTree) was used to select the most significant spectro-textural-morphological leaf signatures resulting in only 11 attributes. These leaf signatures were profiled using the distance method with a distance power of 2. Hybrid CTree and linear discriminant analysis (CTree-LDA or SalviaNet) outperformed other computational models in classifying Salvia species based on the accuracy (90.7%) and sensitivity (90.7%). Based on profiling, leaf's red reflectance, compactness, and shape factor 2 are the strong determinants in discriminating Salvia species. Overall, the developed SalviaNet is proven reliable for on-site application and will essentially help the field of plant taxonomy.

Handwriting recognition poses a huge challenge in image processing due to the lack of defined construction of the characters written by different people. This paper proposes a technique addressing the research gap in manual tuning enhancement of classification of handwritten Arabic Numerals through the introduction of image contouring in the preprocessing stage and its optimization using genetic algorithms. The researchers utilized image contouring techniques in extracting features of the image. The image contouring was assisted by the genetic algorithm in optimizing the tunings of the contours to maximize the features extracted. The experiment utilized the MNIST Datasets for training and testing. Accuracy per character varies due to the number of variations of a given character. Total accuracy of 99% was obtained during the testing.

Data mining today is much slower than before because of the advancement of computing and information systems. Relevant recommendation based on customers' preferences and needs in e-commerce gets more complicated. In the recent pandemic, people are reluctant to go out and has engaged more on internet to get their daily food and services. This phenomenon exacerbated the existing recommendation system, as the data has grown up drastically. In this study, the author recommends a relevant image quality based on the quality queries of the clothes and footwear dataset by observing their highest similarity score. Fashion MNIST images used were existing dataset for clothes and footwear. The testing on image reconstruction using training and validation approaches has shown an accurate result by showing only 0.01 loss in the dataset. Using 11 classes of the image queries, the system image has been identically reconstructed according to the queries supplied. With this result, businesses will have an implementation alternative to a faster and more efficient data mining method. Hence, this alternative will boost the speed of many recommendation systems in the e-commerce platforms and will create a better customer experience.

Excess nitrate concentration leads to excessive algal growth that reduces dissolved oxygen for aquatic animals. A significant strategy to preserve the water quality of aquatic systems is through nitrate level assessment. However, use of nitrate sensors and existing laboratory approach is costly and requires a huge effort. This study investigated the application of computational intelligence for measurement of nitrate concentration in a tilapia fishpond at Rizal province, Philippines, based on physico-limnological parameters such as temperature, electrical conductivity, and pH level. Artificial neural network (ANN) algorithms including feed-forward (FNN) and recurrent (RNN) neural networks were developed and optimized using genetic algorithm (GA) to improve their predicting performances. Genetic programming (GP), through GPTIPSv2 tool, was configured to generate a fitness function. This function is the principal component of GA optimization to produce optimal number of hidden neurons for ANN architecture that resulted in 2 neurons for GA-FNN and combination of 92, 31, and 11 neurons for each hidden layer using the GA-RNN model. Based on evaluation results, all models provided acceptable results with error and predictive accuracy values approaching 0 and 1, respectively. However, the GA-FNN model outperformed other models with 3.26 RMSE, 2.23 MAE, and 0.97 R2values which proved to be the most effective and suitable model for the indirect measurement of nitrate concentration.

Fish feeding management is one of the most crucial considerations in aquaculture production. The traditional feeding method such as table-based and scheduled automated feeding schemes are inaccurate. In contrast, the automated on-demand feeding system has reduced the inaccuracies of the older feeding schemes. However, existing on-demand systems have limited accessibility because their monitoring systems are only stored by their local devices. This paper proposes an on-demand fish feeding system with online and real-time monitoring using the Internet of Things (IoT) and an accelerometer to sense the fish' demand by hitting it. An overhead surveillance camera was installed on the fish tank to automatically record and monitor the fish feeding activity on the first day of the implementation. Two groups of fish were used for the observation—the adults and pre-growth Nile tilapia (Oreochromis niloticus). Results have shown that the on-demand feeding system is highly effective on 21 pre-growth fish with an average weight of 88 grams and a standard deviation (SD) of ± 39 grams. Additionally, the feed intake ratio (FIR) of the pre-growth fish was$1.35\pm 0.69$grams, i.e., 73% to 86% lower than the recommended table-based feeding scheme. Thus, more efficient.

Dehazing through Dark Channel Prior (DCP), originally developed for land-based images, has translated its potential for improving the quality of underwater images. However, the DCP default parameters, which are just adapted from land-based applications, may not be applicable for underwater images. Such constraint limits the capability of this restoration algorithm to improve the quality of an underwater image; the values of these parameters must be searched for each underwater image. A proposed approach on the parameter values assignment problem is to conduct an optimized search based on Genetic Algorithm. The presentation of this proposed approach focuses on the Genetic Algorithm processes: chromosome encoding, fitness function development, and selection, mutation, and crossover, to perform an effective search of the best solution out of a pool of possible solutions. Qualitative and quantitative evaluations show that utilization of optimized combination of DCP parameters, achieves images of higher quality in comparison to the utilization of established default DCP parameters.

With machine learning-based innovations becoming a trend, practical resolutions of its implementation to large-scale data and computing problems must be able to cope up as well. Currently, Graphic Processing Units (GPUs) are being chosen over other available physical devices due to its powerful computing capability and easier handling. Several cloud service providers also made it possible for these to be accessible online allowing higher serviceability and lower cost upfront for businesses. With this said, the proponent would implement a common machine learning-based application, automated annotation through Mask RCNN Object Detection Model in CVAT, using AWS instance. The key purpose is to showcase the viability of deploying data and computing intensive system on the cloud.

Identifying variant diseases in leaves is a significant method for optimizing food production. As the global population continues to arise and agricultural space continues to decline, every possible way of increasing the supply of food in any given condition and limited resources will address the above-mentioned problems. This study proposes a way for detecting three different diseases from grape leaves apart from the healthy leaves and considers the confidence value of the system in correctly identifying the classes. The diseases are namely: Black Rot, Black Measles, and Isariopsis. The system conducted a comparative analysis to determine which among the three pre-trained networks, AlexNet, GoogLeNet, and ResNet-18 will be the most suitable network to be integrated with Regions with Convolutional Neural Networks (RCNN) in performing multiple object detection in a given image. The data used in training the models comprised of annotated image data represented as a ground truth table with image files and their corresponding bounding boxes coordinates. The models evaluated resulted to AlexNet being the best pre-trained network to be working on the RCNN with an accuracy of 95.65%. The other two models from GoogLeNet and ResNet-18 only obtained accuracies of 92.29% and 89.49% respectively.

The technological capabilities of computers in today's time continues to improve in ways that seemed impossible before. It is common knowledge that most people use computers to make everyday lives easier. Therefore, it is vital to bridge the gap between humans and computers to provide more suitable aid to the user. One way to do this is to use emotion recognition as a tool to make the computer understand and analyze how it can help its user on a much deeper level. This paper proposes a way to use both face and speech emotion recognition as a basis for selecting an appropriate music that can improve or relieve one's emotion or stress. To accomplish this, Support Vector Machine with different kernels are used to create the models for validation and testing on both the face and speech emotion recognition. The final integrated system yielded an accuracy rate of 78.5%.

The Aspergillus genus is deemed relevant for distinction and classification in the field of food, agriculture and medicine. As there are harmful and useful ones, it adds to the necessity of correct classification. Categorization of this conidial fungi is usually done through manual microscopical procedures which apparently has a degree of subjectiveness. In order to classify Aspergillus samples faster and more accurately, technology, specifically image processing and machine learning are incorporated in this study. Pre-trained deep learning models are employed in classifying 9 kinds of Aspergillus. The methodology is generally comprised of preprocessing, deep-learning (training) and performance evaluation. Performance evaluation pertains to the validation accuracy and running times of the system after training through visual display of graphs and tabulation of acquired data. This study achieved a 93.3333% testing accuracy proving that the transferred knowledge is accurate, compatible and reliable.

Monitoring the dynamics of fish behavior is impactful both in the research for fisheries and aquaculture production. One of the most explored approaches to monitor the fish is tracking-by-detection along with computer vision. Presently, there are several challenges in this field, including underwater environment conditions and fish movement complexity. This study presents an initial investigation towards tracking the fish by predicting the trajectory 2D coordinates of fish from the sequential sampled frames in underwater videos. Here, the authors explored the Genetic Algorithm based on natural evolution selection and the Long Short-Term Memory (LSTM) algorithm. Results have shown tolerable trajectory prediction inaccuracies using the GA and LSTM. Specifically, it obtained the Mean Absolute Percentage Error at 2.8% to 30.5% and 3.33% to 17.74% for GA and LSTM, respectively. These results have allowed the authors and researchers to extend its study towards tracking the fish using these approaches.

The hybridization of MIMO-OFDM systems became one of the most used wireless communication model for broadband, mobile, and multimedia applications because of its high bandwidth efficiency, bandwidth capacity, and robustness to fading. However, it suffers from the underlying disadvantage of OFDM system which is having a high peak-to-average-power ratio (PAPR) due to large envelope variations. These variations cause non-linear distortion when the OFDM signal is amplified for transmission. Hence, in order to eliminate the non-linear distortion effects of the high power amplifier in MIMO-OFDM systems, the input signal power must have an appropriate power level to satisfy an optimal input back off (IBO) value that also contributes to an amplifier’s maximum efficiency. A Fuzzy Logic Controller is used to control the IBO of the system as well as the signal power level. Results shows that using the proposed Single-Input Single-Output (SISO) Fuzzy Power Controller reduces the bit error rate (BER) significantly compared to the traditional scheme.

The study aims to design an intelligent total body water measuring device which will help to determine the total body water level or percentage of an individual using ultrasonic sensor, load cell and bioelectric impedance analysis (BIA). The system utilized the Scaled Conjugate Gradient Artificial Neural Network (ANN) as the machine learning algorithm. The system used the dataset splitting of 70%-15%15% for training, validation and testing. Different hidden neurons were used and compared during neural network training and found out that using 10 neurons will provide the lowest mean square error (MSE) with best value of Pearson's correlation (R). Based on the results, using 10 neurons, Scaled Conjugate Gradient algorithm has better performance as compared to Levenberg-Marquardt algorithm with MSE equal to 0.180033, 0.118954, 0.529157 while the R value is equal to 0.997887, 0.997488, 0.99644 for training, validation and testing.

Lettuce is one of the most popular crops for urban farming because it is easy to grow and it has high nutritional value. Moreover, it is adaptable and can be combined with other food options, or it can be eaten alone without too much preparation. Predicting lettuce growth can be crucial to find the optimum maturity and harvest time. This paper proposed to use a model of a hybrid tree-fuzzy logic approach, the classification tree was used to select the most significant features from the texture features then the fuzzy inference system was utilized in predicting the lettuce growth stage classification. The hybrid system produced accurate results with low percentage error and correct classifications. Based on these results, the most accurate prediction can be observed in the head development growth stage; the harvest growth stage has a slight variance, while the vegetative stage has the most variance. Overall, the trained hybrid system is reliable in predicting and identifying lettuce growth stage classification.

This paper presents a decentralized and cooperative load lifting and transportation system using unmanned aerial vehicle quadcopters. The limitation of a single UAV to carry load is addressed in this study by creating a cooperative lifting system that can accommodate varying load weight. Cooperative, independent and scalable agents were implemented with decision making algorithm embedded in each agents. Decentralized sensing of load is done by the UAV and the group consensually decides if another UAV is needed to carry the load. The system can lift different weight by autonomously sending appropriate number of UAV depending on the load. Experiments were conducted to determine the responsiveness of the system in varying load weights. Experiment results showed that the developed system is robust and scalable.

In response to addressing the issue of pipe quality checking, the researchers developed an adaptive in-pipe inspection robot that is able to detect rust as well as map the rust on the pipe network. The robot is traversed in a pipe network of horizontal, vertical, elbow, and tee type with diameters of 8, 10 and 12 inches for all. Hence, the test features the versatility, adaptability, and robustness of the robot. The leg expansion of the robot is inspired by the scissors mechanism. On the other hand, rust detection was done through a per pixel classification via image processing. To effectively map the rust, checkpoints were used as a guide of the robot. Testing of the robot were supported in both simulation and actual testing, wherein it yields a 96.45% success rate on the site. Likewise, its rust detection program proved to be successful with a high percentage accuracy of 99.18%. The localization on the other hand yielded an accuracy of 85%. Given the obtained data and results, the researchers were able to go beyond their target objective of 70%.

In this study, a fuzzy logic controller (FLC) was designed to manipulate an articulated robot gripper. An idea from a previous study was utilized to enhance the performance of the FLC using genetic algorithms by optimizing newly-introduced coefficients in the membership functions of the FLC. The proposed controller was applied on a robot gripper model in Simulink. All in all, the genetic algorithm was able to come up with optimized parameters after an average of at least eight (8) generations and the proposed controller was able to follow the reference trajectory more accurately than the simple fuzzy controller. Further research will be necessary for physical implementation and possible improvement of the utilized genetic algorithm.

Vision systems are essential in building a mobile robot that will complete a certain task like navigation, surveillance, and explosive ordnance disposal (EOD). This will make the robot controller or the operator aware what is in the environment and perform the next tasks. With the recent advancement in deep neural networks in image processing, classifying and detecting the object accurately is now possible. In this paper, Convolutional Neural Networks (CNN) is used to detect objects in the environment. Two state of the art models are compared for object detection, Single Shot Multi-Box Detector (SSD) with MobileNetV1 and a Faster Region-based Convolutional Neural Network (Faster-RCNN) with InceptionV2. Result shows that one model is ideal for real-time application because of speed and the other can be used for more accurate object detection.

The property of the Smoothed Particle Hydrodynamics (SPH) method of being mesh free, adaptable and suitable for tracking of individual particles makes it appropriate for approximating swarm behaviors for multi-agent robotics applications. The researchers modeled each of the swarm robots as SPH particles and then subjected them to external forces to exhibit aggregation and force certain formations. The external forces subjected to the SPH particles are gravity forces and container constraints . The containers explored in the study are simple geometrical primitives: sphere and cube . Computer simulations were done to show how SPH can facilitate in forcing swarm formations with the help of bounding primitives. Algorithm benchmarking was done to show how well SPH performs. Results show that SPH performs better than the benchmark algorithm by a margin of 0.703 and 1.016 units for the two set-ups, respectively. Actual robot implementation was also done to verify the effectivity and viability of the proposed algorithm in exhibiting the aggregation behavior. After 15 seconds of system run time, the interparticle distance and motion accuracy reached 96.93% and 91.14%, respectively.