Object Classification Research Articles

FPGA Implementation of Pillar-Based Object Classification for Autonomous Mobile Robot

FPGA Implementation of Pillar-Based Object Classification for Autonomous Mobile Robot

Towards novel classification of infants’ movement patterns supported by computerized video analysis

BackgroundPositional preferences, asymmetry of body position and movements potentially indicate abnormal clinical conditions in infants. However, a lack of standardized nomenclature hinders accurate assessment and documentation of these preferences over time. Video tools offer a safe and reproducible method to analyze and describe infant movement patterns, aiding in physiotherapy management and goal planning. The study aimed to develop an objective classification system for infant movement patterns with particular emphasis on the specific distribution of muscle tension, using methods of computer analysis of video recordings to enhance accuracy and reproducibility in assessments.MethodsThe study involved the recording of videos of 51 infants between 6 and 15 weeks of age, born at term, with an Apgar score of at least 8 points. Based on observations of a recording of infant spontaneous movements in the supine position, experts identified postural-motor patterns: symmetry and typical asymmetry linked to the asymmetrical tonic neck reflex. Deviations from the typical postural-motor system were indicated, and subcategories of atypical patterns were distinguished. A computer-based inference system was developed to automatically classify individual patterns.ResultsThe following division of motor patterns was used: (1) normal patterns, including (a) typical (symmetrical, asymmetrical: variants 1 and 2); and (b) atypical (variants: 1 to 4), (2) positional preference, and (3) abnormal patterns. The proposed automatic classification method achieved an expert decision mapping accuracy of 84%. For atypical patterns, the high reproducibility of the system’s results was confirmed. Lower reproducibility, not exceeding 70%, was achieved with typical patterns.ConclusionsBased on the observation of infant spontaneous movements, it is possible to identify movement patterns divided into typical and atypical patterns. Computer-based analysis of infant movement patterns makes it possible to objectify and satisfactorily reproduce diagnostic decisions.

Severe Weather Warnings for Sofia, Bulgaria: May – September 2010–2019

Severe weather monitoring and forecasting is the prime task of the National Meteorological Services according to World Meteorological Organization Guidelines (WMO). Since 2009 Bulgarian National Institute of Meteorology and Hydrology (NIMH) issues 48 h ahead severe weather warnings as a contribution to the European warning system (METEOALARM). A total of 243 days with warnings for heat, thunderstorms and intense rain were issued for the period May-September 2010–2019. The monthly maximum for heat is 42 days in July, for thunderstorms it is 26 days in June, and for intense rain it is 27 days in June. Objective circulation classification is conducted which shows that the majority of heat warnings are associated with Anticyclonic and West flow circulation types. About 40% of thunderstorm warnings are of Cyclone type followed by 28.6% of Cyclonic Directional types. For rain warnings, most of the days are with Cyclone type (46.2%) and Cyclonic Directional types (27.5%). Analysis of Global Navigation Satellite Systems (GNSS) derived Integrated Water Vapour (IWV), from the Bulgarian station in Sofia, shows consistently higher values for the days with METEOALARM warnings for heat, thunderstorm and rain.

OPTICAL DEEP LEARNING LANDMINE DETECTION BASED ON LIMITED DATASET OF AERIAL IMAGERY

Landmine detection is one of the most innovative applications of unmanned aerial vehicles that became possible due to rapid development of both aerial vehicles equipped by different optical cameras and sensors using different physical principles, and object classification and detection methods, including machine learning and especially deep learning. Optical detection is an essential part of the overall landmine detection process that can be performed either separately or in combination with data processing from other types of cameras or sensors. The development of deep convolutional neural networks has dramatically changed the landscape of optical detection by making them de-facto choice number one for the majority of object classification, detection and segmentation tasks. However, the deterrent factor in the case of landmine detection is limited availability of appropriate data for training that different researchers try to overcome in different ways. The assessment of necessary amount of training data for any particular object detection problem still remains an experimental task. Despite several years of development in this area, still there is a shortage of research based on real landmine imagery obtained from unmanned aerial vehicles, so currently any public effort in this direction is valuable and works as an inspiration for new researchers. This paper describes such a study, namely its first iteration in which popular open-source tools are used to build detection pipeline and estimation of their efficiency is done using limited amount of data. It is shown that the problem of limited amount of training data can be effectively overcome by data augmentation and iterational process of training optical landmine detector is demonstrated. The effectiveness of open-source tools and libraries for neural networks training, object detection and dataset preparation is also demonstrated.

Design of AI-Enhanced and Hardware-Supported Multimodal E-Skin for Environmental Object Recognition and Wireless Toxic Gas Alarm

HighlightsA novel organohydrogel-based multimodal e-skin with excellent sensing performance for temperature, humidity, pressure, proximity, and NO2 is proposed for the first time, showing powerful sensing capabilities beyond natural skin.The developed multimodal e-skin exhibited extraordinary sensing performance at room temperature, including fast pressure response time (0.2 s), high temperature sensitivity (9.38% °C-1), a wide range of humidity response (22%–98% RH), high NO2 sensitivity (254% ppm-1), a low detection limit (11.1 ppb NO2) and the abilities to sense the proximity of objects accurately, which are yet achieved by previous e-skins.The multimodal e-skin was combined with the deep neural network algorithm and wireless alarm circuit to achieve zero-error classification of different objects and rapid response to NOx leak incidents, proving the feasibility of the e-skin-assisted rescue robot for post-earthquake rescue.

Advancing 3D point cloud understanding through deep transfer learning: A comprehensive survey

The 3D point cloud (3DPC) has significantly evolved and benefited from the advances of deep learning (DL). However, the latter faces various issues, including the lack of data or annotated data, the existence of a significant gap between training data and test data, and the requirement for high computational resources. To that end, deep transfer learning (DTL), which decreases dependency and costs by utilizing knowledge gained from a source data/task in training a target data/task, has been widely investigated. Numerous DTL frameworks have been suggested for aligning point clouds obtained from several scans of the same scene. Additionally, DA, which is a subset of DTL, has been modified to enhance the point cloud data’s quality by dealing with noise and missing points. Ultimately, fine-tuning and DA approaches have demonstrated their effectiveness in addressing the distinct difficulties inherent in point cloud data. This paper presents the first review shedding light on this aspect. it provides a comprehensive overview of the latest techniques for understanding 3DPC using DTL and domain adaptation (DA). Accordingly, DTL’s background is first presented along with the datasets and evaluation metrics. A well-defined taxonomy is introduced, and detailed comparisons are presented, considering different aspects such as different knowledge transfer strategies, and performance. The paper covers various applications, such as 3DPC object detection, semantic labeling, segmentation, classification, registration, downsampling/upsampling, and denoising. Furthermore, the article discusses the advantages and limitations of the presented frameworks, identifies open challenges, and suggests potential research directions.

A new CNN-BASED object detection system for autonomous mobile robots based on real-world vehicle datasets

Recently, autonomous mobile robots (AMRs) have begun to be used in the delivery of goods, but one of the biggest challenges faced in this field is the navigation system that guides a robot to its destination. The navigation system must be able to identify objects in the robot's path and take evasive actions to avoid them. Developing an object detection system for an AMR requires a deep learning model that is able to achieve a high level of accuracy, with fast inference times, and a model with a compact size that can be run on embedded control systems. Consequently, object recognition requires a convolutional neural network (CNN)-based model that can yield high object classification accuracy and process data quickly. This paper introduces a new CNN-based object detection system for an AMR that employs real-world vehicle datasets. First, we create original real-world datasets of images from Banda Aceh city. We then develop a new CNN-based object identification system that is capable of identifying cars, motorcycles, people, and rickshaws under morning, afternoon, and evening lighting conditions. An SSD Mobilenetv2 FPN Lite 320 × 320 architecture is employed for retraining using these real-world datasets. Quantitative and qualitative performance indicators are then applied to evaluate the CNN model. Training the pre-trained SSD Mobilenetv2 FPN Lite 320 × 320 model improves its classification and detection accuracy, as indicated by its performance results. We conclude that the proposed CNN-based object detection system has the potential for use in an AMR.

Six-Axis Robotic Gripping Control Algorithm Based on Deep Learning

Due to several complex factors such as the type, size, and shape of target object, vision-assisted robot grasping technology still faces serious challenges. In this research, a deep learning-based robot hand vision grasping algorithm was developed considering semi-structural environmental constraints. The proposed algorithm could build a deep learning network on the basis of the desired object, perform object recognition, category classification and position judgment, and complete robot hand-grasping tasks. The obtained experimental results demonstrated that the proposed algorithm effectively solved the problem of recognizing and classifying multi-category objects in a semi-structured environment, improving recognition rate and grasping rate and reducing collision rate.

Unpaired Remote Sensing Image Dehazing Using Enhanced Skip Attention-Based Generative Adversarial Networks with Rotation Invariance

Remote sensing image dehazing aims to enhance the visibility of hazy images and improve the quality of remote sensing imagery, which is essential for various applications such as object detection and classification. However, the lack of paired data in remote sensing image dehazing enhances the applications of unpaired image-to-image translation methods. Nonetheless, the considerable parameter size of such methods often leads to prolonged training times and substantial resource consumption. In this work, we propose SPRGAN, a novel approach leveraging Enhanced Perlin Noise-Based Generative Adversarial Networks (GANs) with Rotation Invariance to address these challenges. Firstly, we introduce a Spatial-Spectrum Attention (SSA) mechanism with Skip-Attention (SKIPAT) to enhance the model’s ability to interpret and process spectral information in hazy images. Additionally, we have significantly reduced computational overhead to streamline processing. Secondly, our approach combines Perlin Noise Masks in pre-training to simulate real foggy conditions, thereby accelerating convergence and enhancing performance. Then, we introduce a Rotation Loss (RT Loss) to ensure the model’s ability to dehaze images from different angles uniformly, thus enhancing its robustness and adaptability to diverse scenarios. At last, experimental results demonstrate the effectiveness of SPRGAN in remote sensing image dehazing, achieving better performance compared to state-of-the-art methods.

Fish Detection and Classification using YOLOv8 for Automated Sorting Systems

Automation plays a crucial role in scaling up freshwater fish cultivation to address the future threat of food scarcity and meet growing nutrition needs. The fish industry, in particular, develops automation in the sorting and selection processes. However, research on this technology's development is still very limited. In this work, we propose an approach for detecting and classifying fish running on conveyors. We use YOLOv8, which is the most popular and newest deep learning model for object detection and classification. We conducted our test using the KMITLFish dataset, a moving conveyor belt recording that encompasses common cultivated freshwater fish in Thailand along with some endemic species. As a result, our proposed method was able to accurately detect and classify eight types of fish at a conveyor speed of 505.08 m/h. Moreover, we developed this work using a ready-to-use AI platform, intending to directly contribute to the advancement of automatic fish sorting system technology in the fish industry.

Object classification based on impulse response analysis using vibration propagation

Object classification based on impulse response analysis using vibration propagation

Non-Invasive Raman Classification Comparison with pXRF of Monochrome and Related Qing Porcelains: Lead-Rich-, Lead-Poor-, and Alkali-Based Glazes.

Chinese porcelain with an optically clear colored glaze, imported to Europe from the Kangxi period (1662-1722, Qing Dynasty) onwards was highly collected by the French Elite of the 18th century. The bright colors with a clear, shiny glaze were unlike anything produced in Europe at that time. The colors of enamelled artifacts (on biscuits or already glazed porcelain) can be fully monochrome or consist of associations of large monochromatic areas with or without application of gilding. Non-invasive portable XRF and mobile Raman analyses have previously shown their effectiveness in the characterization of (colored) glassy silicates. In this study, we compare the Raman signatures of twenty-one Chinese artifacts fully-or with major monochrome area (sancai)-decorated with blue, turquoise (or celectian blue), honey-yellow, green, eggplant, and red color. Different types of glazes are identified and confirmed by pXRF: lead-rich, lead-poor-alkali, lead-doped alkali, and alkali-based compositions. However, an unexpected low level of lead is observed in the turquoise glazes, likely to optimize the gloss. Raman spectroscopy appears more reliable to compare the Pb content than pXRF. This work presents Raman spectral signatures of glazes that can potentially be used for non-invasive object classification and counterfeit detection.

Intelligent Surveillance of Airport Apron: Detection and Location of Abnormal Behavior in Typical Non-Cooperative Human Objects

Most airport surface surveillance systems focus on monitoring and commanding cooperative objects (vehicles) while neglecting the location and detection of non-cooperative objects (humans). Abnormal behavior by non-cooperative objects poses a potential threat to airport security. This study collects surveillance video data from civil aviation airports in several regions of China, and a non-cooperative abnormal behavior localization and detection framework (NC-ABLD) is established. As the focus of this paper, the proposed framework seamlessly integrates a multi-scale non-cooperative object localization module, a human keypoint detection module, and a behavioral classification module. The framework uses a serial structure, with multiple modules working in concert to achieve precise position, human keypoints, and behavioral classification of non-cooperative objects in the airport field. In addition, since there is no publicly available rich dataset of airport aprons, we propose a dataset called IIAR-30, which consists of 1736 images of airport surfaces and 506 video clips in six frequently occurring behavioral categories. The results of experiments conducted on the IIAR-30 dataset show that the framework performs well compared to mainstream behavior recognition methods and achieves fine-grained localization and refined class detection of typical non-cooperative human abnormal behavior on airport apron surfaces.

SRGAN-enhanced unsafe operation detection and classification of heavy construction machinery using cascade learning

In the inherently hazardous construction industry, where injuries are frequent, the unsafe operation of heavy construction machinery significantly contributes to the injury and accident rates. To reduce these risks, this study introduces a novel framework for detecting and classifying these unsafe operations for five types of construction machinery. Utilizing a cascade learning architecture, the approach employs a Super-Resolution Generative Adversarial Network (SRGAN), Real-Time Detection Transformers (RT-DETR), self-DIstillation with NO labels (DINOv2), and Dilated Neighborhood Attention Transformer (DiNAT) models. The study focuses on enhancing the detection and classification of unsafe operations in construction machinery through upscaling low-resolution surveillance footage and creating detailed high-resolution inputs for the RT-DETR model. This enhancement, by leveraging temporal information, significantly improves object detection and classification accuracy. The performance of the cascaded pipeline yielded an average detection and first-level classification precision of 96%, a second-level classification accuracy of 98.83%, and a third-level classification accuracy of 98.25%, among other metrics. The cascaded integration of these models presents a well-rounded solution for near-real-time surveillance in dynamic construction environments, advancing surveillance technologies and significantly contributing to safety management within the industry.

Self-driving car implementation guided by computer vision for detection

This conference paper discusses the influence of computer vision in self-driving car implementation. The automotive industry has experienced a significant transformation by integrating artificial intelligence, such as computer vision. The global market for AI in automotive has grown remarkably since many organizations in the electric vehicle industry have engaged in computer vision due to its benefits. Object detection in computer vision contributes to autonomous driving systems, assisting self-driving vehicles in classification of the road objects accurately. This paper has provided an overview of the importance of computer vision to self-driving cars, such as user experience, reduction in cost and vehicle safety. However, some challenges, such as adverse weather conditions, have also been experienced. A case study analysis has been done to have practical implications of the self-driving car implementation guided by computer vision for detection.

Radar-Based Pedestrian and Vehicle Detection and Identification for Driving Assistance

The introduction of advanced driver assistance systems has significantly reduced vehicle accidents by providing crucial support for high-speed driving and alerting drivers to imminent dangers. Despite these advancements, current systems still depend on the driver’s ability to respond to warnings effectively. To address this limitation, this research focused on developing a neural network model for the automatic detection and classification of objects in front of a vehicle, including pedestrians and other vehicles, using radar technology. Radar sensors were employed to detect objects by measuring the distance to the object and analyzing the power of the reflected signals to determine the type of object detected. Experimental tests were conducted to evaluate the performance of the radar-based system under various driving conditions, assessing its accuracy in detecting and classifying different objects. The proposed neural network model achieved a high accuracy rate, correctly identifying approximately 91% of objects in the test scenarios. The results demonstrate that this model can be used to inform drivers of potential hazards or to initiate autonomous braking and steering maneuvers to prevent collisions. This research contributes to the development of more effective safety features for vehicles, enhancing the overall effectiveness of driver assistance systems and paving the way for future advancements in autonomous driving technology.

Deep Learning-based Multi-class Three-dimensional (3-D) Object Classification using Phase-only Digital Holographic Information

In this paper, we present a deep CNN-based approach for multi-class classification of three-dimensional (3-D) objects using phase-only digital holographic information. The 3-D objects considered for the multi-class (four-class) classification task are ‘triangle-square’, ‘circle-square’, ‘square-triangle’, and ‘triangle-circle’. The 3-D object ‘triangle-square’ is considered for Class-1 and the remaining 3-D objects ‘circle-square’, ‘square-circle’, and ‘triangle-circle’ are considered for Class-2, Class-3, and Class-4. The digital holograms of 3-D objects were created using the two-step Phase-Shifting Digital Holography (PSDH) technique and were computationally post-processed to obtain phase-only digital holographic data. Subsequently, a deep CNN was trained on a phase-only image dataset consisting of 2880 images to produce the results. The loss and accuracy curves are presented to validate the performance of the model. Additionally, the results are validated using metrics such as the confusion matrix, classification report, Receiver Operating Characteristic (ROC) curve, and precision-recall curve.

Vision transformer and CNN-based skin lesion analysis: classification of monkeypox

AbstractMonkeypox is an important health problem. Rapid diagnosis of monkeypox skin lesions and emergency isolation when necessary is essential. Also, some skin lesions, such as melanoma, can be fatal and must be rapidly distinguished. However, in some cases, it is difficult to distinguish the lesions visually. Methods such as dermoscopy, high-resolution ultrasound imaging, etc. can be used for better observation. But these methods are often based on qualitative analysis, subjective and time-consuming. Therefore, in this study, a quantitative and objective classification tool has been developed to assist dermatologists and scientists. The proposed system classifies seven skin lesions, including monkeypox. A popular approach Vision Transformer and some popular deep learning convolutional networks have been trained with the transfer learning approach and all results have been compared. Then, the models that show the best accuracy score have been combined to make the final prediction using bagging-ensemble learning. The proposed ensemble-based system produced 81.91% Accuracy, 65.94% Jaccard, 87.16% Precision, 74.12% Recall, and 78.16% Fscore values. In terms of different criteria metrics, the system produced competitive or even better results than the literature.

Review and Classification of Objectives in Dynamic Dial-a-Ride Systems: A Triple Bottom Line Approach of Sustainability

Dynamic dial-a-ride problems (DDARPs) involve designing routes and schedules for customers with specific origins and destinations. While the optimization of DDARPs has been extensively examined, these analyses often focus solely on economic decisions. The recent literature emphasizes the inclusion of social and environmental factors in addition to economic considerations for a sustainable transportation system. This paper provides a conceptual review that identifies and classifies the most common DDARP objectives in the three dimensions of the Triple-Bottom-Line (3BL) approach of sustainability: environmental, economic, and social. This study analyzes the interconnections among different objectives and provides insights into multi-objective approaches used in transportation problems. The findings demonstrate the interconnectedness of objectives from different dimensions and highlight the involvement of various stakeholders in decision-making. The results show that optimizing one objective may have implications for other objectives, suggesting a trade-off to be considered. The results reveal that social objectives boost the economic dimension by improving service quality; however, environmental objectives negatively impact the economic dimension. Additionally, a geographical analysis was conducted, which revealed continent-wise variations in research focus and contributions. Future studies should focus more on the social and environmental dimensions to promote a sustainable transportation system.

Feasibility study on the use of peripheral blood differentially expressed genes for objective classification of chronic subjective tinnitus: a case study on high-frequency tinnitus

Objective: To explore the feasibility of constructing an objective tinnitus subtype model based on peripheral blood differentially expressed genes (DEGs) using a combination of Weighted Gene Co-expression Network Analysis (WGCNA) and Random Forest algorithm (RF). Methods: From October 2019 to June 2020, peripheral blood DEGs were obtained from 37 patients (from the Third Affiliated Hospital of Sun Yat-sen University)with chronic subjective high-frequency tinnitus (21 unbothersome type, 16 bothersome type) and 20 healthy volunteers through high-throughput sequencing. WGCNA was used to construct gene modules with different expression patterns and analyze their relationships with tinnitus characteristics. Subsequently, RF was employed to build subtype models, which were evaluated by the area under the receiver operating characteristic curve (AUC), accuracy, and F1-score. Results: A total of 12 351 intergroup DEGs were divided into 9 gene modules. Among them, MEblue, MEgreen, and MEbrown showed significant negative correlations with the healthy volunteer group, while MEpink showed a significant positive correlation with the tinnitus distress group. The "Tinnitus vs. Normal" and "Compensatory vs. Decompensatory" subtype models, based on MEblue and MEpink respectively, both had AUCs greater than 0.80, accuracies above 90%, and F1-scores above 0.90, indicating good performance. Conclusions: Peripheral blood DEGs are potential biological indicators for objective classification of subjective tinnitus. The combined application of WGCNA and the Random Forest algorithm should be a viable approach to constructing an objective tinnitus subtype model. However, further exploration and refinement are needed to validate the model's generalizability, cross-dataset performance, and algorithm optimization.