International Journal of Advances in Applied Computational Intelligence
Journal DOI
2833-5600ISSN (Online)
Object detection in remote sensing images (RSI) is a main procedure where the purpose is to automatically recognize and categorize certain objects or features from large-scale, remotely developed images like aerial imagery or satellite. This task role a vital play in extracting appreciated data from massive geographical regions, contributing to various applications under several domains namely environmental monitoring, urban planning, agriculture, and disaster management. Recent developments in deep learning (DL) technologies have significantly enhanced the accuracy and efficacy of object detection systems for RS, enabling more precise and automated analysis of various landscapes and facilitating informed decision-making. DL approaches namely convolutional neural networks (CNNs) are exposed to remarkable abilities in learning intricate patterns and features from difficult spatial data, resulting in enhanced accuracy and effectiveness. In this article, we present a Towards Efficient Hyperspectral Object Detection and Classification using Thermal Optimization Algorithm with Deep Learning (HODC-TOADL) system. The objective of HODC-TOADL algorithm is to identify and categorize distinct types of objects that exist in the RSI. In the HODC-TOADL method, an improved Dense Net model is applied to learn the distinct features of the input RSI. Besides, the TOA has been deployed to boost the hyper parameter choice of the Dense Net method. Furthermore, the classification of objects can be carried out by employing of adaptive neurofuzzy inference system (ANFIS). The experimental evaluation of the HODC-TOADL algorithm can be studied on benchmark databases. The experimental values stated that the HODC-TOADL algorithm reaches effective classification performance compared to recent DL models.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060201
Vol. 6 Issue. 2 PP. 1-15, (2024)
In recent decades, Red Palm Weevils (RPW) have been demonstrated as a harmful pest of palm trees worldwide, predominantly in the Middle East. The RPW is produced massive damage to several palm varieties. Primary detection of the RPW is a complex problem to optimum date production while the recognition is avoided by palm trees as to be influenced by RPW. Several studies are driven to determine a precise approach for the detection, localization, and classification of RPW pests. Employing computer vision (CV) technology with pattern detection is verified that further productive once utilized for identifying and classifying insects. Thus, the automated method decreases either the problem or labor effort required for enhancing the farmer's income. The farmers can be stimulated to enhance the productivity of date fruit once this has been done. With this motivation, this article focuses on the design of automated RPW pest detection using sparrow search optimization with deep learning (RPWPD-SSODL) technique. The presented RPWPD-SSODL algorithm mostly focused on the detection and classification of RPW using computer vision approaches. To accomplish this, the RPWPD-SSODL technique employs bilateral filtering (BF) for noise removal. Next, the RPWPD-SSODL technique uses Dense-RefineDet object detector with ShuffleNet model as a backbone network. For improving the recognition solution, the hyperparameter tuning of the ShuffleNet model can be optimally adjusted using the SSO algorithm. To validate the simulation results of the RPWPD-SSODL technique, a wide-ranging simulation outcome is implemented. The simulation values potrayed the improvement of the RPWPD-SSODL algorithm over other approaches under several measures.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060202
Vol. 6 Issue. 2 PP. 16-27, (2024)
This paper is concerned with the study of some novel techniques that using artificial intelligence to protect networks of CAVs from cyberattacks, where we use some machine learning algorithms to detect attacks and compare the machine learning algorithms used for this in terms of accuracy and required operating time. Also, WEKA tool will be used for the desired comparison, as the experiments are carried out on a new dataset, which is a dataset abbreviated from the KDD99 dataset.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060203
Vol. 6 Issue. 2 PP. 28-36, (2024)
This paper is concerned with studying the matrix computations of 3-cyclic refined neutrosophic matrices and 4-cyclic refined neutrosophic matrices with 3cyclic/4-cyclic real entries, where we introduce a novel method to compute eigenvalues and vectors of these matrix classes. Also, we provide a novel algorithm for diagonalization these matrices and to determine whether an n-cyclic refined matrix is diagonalizable or not for n=3, 4.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060204
Vol. 6 Issue. 2 PP. 37-45, (2024)
Recently, computer vision, unmanned aerial vehicles (UAV) based remote sensing (RS) and deep learning (DL) technologies have been instrumental in global food productivity and future agriculture. UAV provides several advantages over other possible RS platforms like real-time data acquisition, high flexibility, and the best tradeoff between spatial, low cost, small size, spectral, and temporal resolution. One possible advantage of using UAVs for crop classification is that they can efficiently and quickly cover large areas, and could gather data from different angles and at different times. This might assist in providing detailed knowledge of the crops and their conditions. Earlier research is limited to finding a single crop from the RGB images taken by the UAV and hasn’t explored the possibility of multi-crop classification by carrying out DL algorithms. Thus, this study presents a new Automated Crop Type Classification using Adaptive African Vulture Optimization with Deep Learning (ACCT-AAVODL) technique. The ACCT-AAVODL algorithm aims to investigate the UAV images and determine different types of food crops. To accomplish this, the presented ACCT-AAVODL method uses a densely connected network (DenseNet121) for generating feature vectors. Since the trial and error hyper parameter tuning is a challenging task, the AAVO model is employed for hyper parameter optimization. The ACCT-AAVODL technique involves a sparse auto encoder (SAE) with a Nadam optimizer for crop type classification, the stimulation analysis of the ACCT-AAVODL approach on the drone imagery dataset shows the remarkable performance of the ACCT-AAVODL method over other approaches.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060205
Vol. 6 Issue. 2 PP. 46-61, (2024)
Examining the temporal behavior of common patterns, obtaining appropriate clusters, and reducing the size of discovered patterns are three significant challenges in temporal data mining. Among the available methods, the constraint-based pattern mining approach has achieved remarkable progress in this domain. Apriori and Interleaved algorithms, which are both slow and outdated, are nonetheless used by present time-granularity pattern exploration approaches. To address these issues, we propose the Frequent Pattern Growth method with Special Constraints. The system incorporates a method for generating patterns on a regular basis. It mandates that transactional datasets adhere to complete and partial cyclic criteria. To locate all possible periodic patterns within the Spatio temporal database, we redefine the task as periodic pattern mining in this thesis. The proposed method makes use of a periodic pattern tree miner. To begin, the clustering method uses an innovative global pollination artificial fish swarm technique to create the most effective dense clusters.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060206
Vol. 6 Issue. 2 PP. 62-72, (2024)
A capability that is indispensable in robotic navigation when it comes to planning paths through dynamic and uncertain environments efficiently and accurately. This work aims at a hybrid stochastic-deterministic path planning by combining the best of both worlds in order to improve robotic navigation. This hybrid model uses stochastic techniques to employ the robustness of uncertainly models, but offers efficient execution with deterministic algorithms for our optimum path solution. The method combines a highly exploratory stochastic sampling-based planner for environmental search with a deterministic optimization component that refines paths generated by the former, enforcing constraints such as minimal traversal distance (energy efficiency), while avoiding obstacles. The integration of those methods targets to override the disadvantages that each purely stochastic or solely deterministic model required, giving a more flexible and robust solution for autonomous vehicle guidance. We use simulation analysis and real-life experimental data to validate the algorithm in comparison with traditional algorithms. The approach performs significantly better, up to an order of magnitude in terms of accuracy and efficiency on navigation as well as robustness against cluttered or dynamic disturbances. These results indicate that the proposed hybrid stochastic-deterministic path-planning algorithm has strong potential to contribute to improving autonomy of robotic navigation systems, especially in highly dynamic and precise applications. The post provides a new framework to improve autonomous navigation of robots for complex environments that can support more efficient, reliable and high-level robotic systems in industrial, household or exploratory settings.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060207
Vol. 6 Issue. 2 PP. 73-82, (2024)
Software defect prediction is a technique that may foretell when and where software errors will manifest. It should be the aim of every software development project to provide a product devoid of bugs. Software defect prediction (SDP) is a crucial aspect of software repair that involves predicting potential code locations for problems. Software of excellent quality need to be bug-free. Software metrics are assessments of the program or its needs that are either quantitative or qualitative in nature. The Lévy flying patterns of various birds and fruit flies, together with the flight patterns of some cuckoo species, served as inspiration for Cuckoo Search (CS), a population-based algorithm that was developed relatively recently. Computer science satisfies the requirements for global convergence. Among the many supervised learning methods that do not need parameters, KNN stands out. This study provides a social metaphorical overview of Stochastic Diffusion Search (SDS) to show how SDS distributes resources. Using a new probabilistic approach, SDS solved the problems of best-fit pattern recognition and matching. Using interactions amongst basic agents, SDS is a distributed computing paradigm that employs multiagent population-based global search and optimization. An optimization approach that combines CS and SDS methods is suggested in this work. This hybridization proposal seeks to improve the cuckoo bird's search strategy for the ideal host nest by using the global search strategy solution of the SDS algorithm. So, to find the best spot for the cuckoo egg, the SDS approach would be used. One possible explanation for PC2's superior performance when compared to other classifiers is its greater recall values. Specifically, KNN outperforms Radial Bias Neural Network (2.20% improvement) and Naive Bayes (7.54% improvement) classifiers.
Read MoreDoi: https://doi.org/10.54216/IJAACI.060208
Vol. 6 Issue. 2 PP. 83-91, (2024)