A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS

Machine Learning and Knowledge Extraction, Journal Year: 2023, Volume and Issue: 5(4), P. 1680 - 1716

Published: Nov. 20, 2023

YOLO has become a central real-time object detection system for robotics, driverless cars, and video monitoring applications. We present comprehensive analysis of YOLO’s evolution, examining the innovations contributions in each iteration from original up to YOLOv8, YOLO-NAS, with transformers. start by describing standard metrics postprocessing; then, we discuss major changes network architecture training tricks model. Finally, summarize essential lessons development provide perspective on its future, highlighting potential research directions enhance systems.

Language: Английский

---

DenseSPH-YOLOv5: An automated damage detection model based on DenseNet and Swin-Transformer prediction head-enabled YOLOv5 with attention mechanism

Advanced Engineering Informatics, Journal Year: 2023, Volume and Issue: 56, P. 102007 - 102007

Published: April 1, 2023

Language: Английский

---

A novel deep learning-based approach for malware detection

Engineering Applications of Artificial Intelligence, Journal Year: 2023, Volume and Issue: 122, P. 106030 - 106030

Published: March 9, 2023

Language: Английский

---

YOLOv1 to v8: Unveiling Each Variant–A Comprehensive Review of YOLO

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 42816 - 42833

Published: Jan. 1, 2024

This paper implements a systematic methodological approach to review the evolution of YOLO variants. Each variant is dissected by examining its internal architectural composition, providing thorough understanding structural components. Subsequently, highlights key innovations introduced in each variant, shedding light on incremental refinements. The includes benchmarked performance metrics, offering quantitative measure variant's capabilities. further presents variants across diverse range domains, manifesting their real-world impact. structured ensures comprehensive examination YOLOs journey, methodically communicating advancements and before delving into domain applications. It envisioned, incorporation concepts such as federated learning can introduce collaborative training paradigm, where models benefit from multiple edge devices, enhancing privacy, adaptability, generalisation.

Language: Английский

---

An efficient and robust Phonocardiography (PCG)-based Valvular Heart Diseases (VHD) detection framework using Vision Transformer (ViT)

Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 158, P. 106734 - 106734

Published: March 1, 2023

Language: Английский

---

Deep learning-accelerated computational framework based on Physics Informed Neural Network for the solution of linear elasticity

Neural Networks, Journal Year: 2023, Volume and Issue: 162, P. 472 - 489

Published: March 13, 2023

Language: Английский

---

A data-driven physics-constrained deep learning computational framework for solving von Mises plasticity

Engineering Applications of Artificial Intelligence, Journal Year: 2023, Volume and Issue: 122, P. 106049 - 106049

Published: March 7, 2023

Language: Английский

---

An Improved YOLOv8 to Detect Moving Objects

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 59782 - 59806

Published: Jan. 1, 2024

Deep learning has revolutionized object detection, with YOLO (You Only Look Once) leading in real-time accuracy. However, detecting moving objects visual streams presents distinct challenges. This paper proposes a refined YOLOv8 detection model, emphasizing motion-specific detections varied contexts. Through tailored preprocessing and architectural adjustments, we heighten the model's sensitivity to movements. Rigorous testing against KITTI, LASIESTA, PESMOD, MOCS benchmark datasets revealed that modified outperforms state-of-the-art models, especially environments significant movement. Specifically, our model achieved an accuracy of 90%, mean Average Precision (mAP) maintained processing speed 30 frames per second (FPS), Intersection over Union (IoU) score 80%. offers detailed insight into trajectories, proving invaluable areas like security, traffic management, film analysis where motion understanding is critical. As importance dynamic scene interpretation grows artificial intelligence computer vision, proposed enhanced highlights potential specialized underscores significance findings evolving field detection.

Language: Английский

---

Detection of Soybean Insect Pest and a Forecasting Platform Using Deep Learning with Unmanned Ground Vehicles

Agronomy, Journal Year: 2023, Volume and Issue: 13(2), P. 477 - 477

Published: Feb. 6, 2023

Soybeans (Glycine max (L.) Merr.), a popular food resource worldwide, have various uses throughout the industry, from everyday foods and health functional to cosmetics. are vulnerable pests such as stink bugs, beetles, mites, moths, which reduce yields. Riptortus pedestris (R. pedestris) has been reported cause damage pods leaves soybean growing season. In this study, an experiment was conducted detect R. according three different environmental conditions (pod filling stage, maturity artificial cage) by developing surveillance platform based on unmanned ground vehicle (UGV) GoPro CAM. Deep learning technology (MRCNN, YOLOv3, Detectron2)-based models used in can be quickly challenged (i.e., built with lightweight parameter) immediately through web application. The image dataset distributed random selection for training, validation, testing then preprocessed labeling annotation. deep model localized classified individuals bounding box masking data. achieved high performances, at 0.952, 0.716, 0.873, respectively, represented calculated means of average precision (mAP) value. manufactured will enable identification field effective tool insect forecasting early stage pest outbreaks crop production.

Language: Английский

---

Pedestrian Detection and Tracking System Based on Deep-SORT, YOLOv5, and New Data Association Metrics

Information, Journal Year: 2023, Volume and Issue: 14(4), P. 218 - 218

Published: April 3, 2023

Pedestrian tracking and detection have become critical aspects of advanced driver assistance systems (ADASs), due to their academic commercial potential. Their objective is locate various pedestrians in videos assign them unique identities. The data association task problematic, particularly when dealing with inter-pedestrian occlusion. This occurs multiple cross paths or move too close together, making it difficult for the system identify track individual pedestrians. Inaccurate can lead false alarms, missed detections, incorrect decisions. To overcome this challenge, our paper focuses on improving pedestrian system’s Deep-SORT algorithm, which solved as a linear optimization problem using newly generated cost matrix. We introduce set new matrices that rely metrics such intersections, distances, bounding boxes. evaluate trackers real time, we use YOLOv5 images. also perform experimental evaluations Multiple Object Tracking 17 (MOT17) challenge dataset. proposed demonstrate promising results, showing an improvement most MOT performance compared default intersection over union (IOU)

Language: Английский

---