YOLOv8: A Novel Object Detection Algorithm with Enhanced Performance and Robustness

Published: April 18, 2024

In recent years, the You Only Look Once (YOLO) series of object detection algorithms have garnered significant attention for their speed and accuracy in real-time applications. This paper presents YOLOv8, a novel algorithm that builds upon advancements previous iterations, aiming to further enhance performance robustness. Inspired by evolution YOLO architectures from YOLOv1 YOLOv7, as well insights comparative analyses models like YOLOv5 YOLOv6, YOLOv8 incorporates key innovations achieve optimal accuracy. Leveraging mechanisms dynamic convolution, introduces improvements specifically tailored small detection, addressing challenges highlighted YOLOv7. Additionally, integration voice recognition techniques enhances algorithm's capabilities video-based demonstrated The proposed undergoes rigorous evaluation against state-of-the-art benchmarks, showcasing superior terms both computational efficiency. Experimental results on various datasets confirm effectiveness across diverse scenarios, validating its suitability real-world contributes ongoing research presenting versatile high-performing algorithm, poised address evolving needs computer vision systems.

Language: Английский

---

A Comprehensive Systematic Review of YOLO for Medical Object Detection (2018 to 2023)

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 57815 - 57836

Published: Jan. 1, 2024

YOLO (You Only Look Once) is an extensively utilized object detection algorithm that has found applications in various medical tasks. This been accompanied by the emergence of numerous novel variants recent years, such as YOLOv7 and YOLOv8. study encompasses a systematic exploration PubMed database to identify peer-reviewed articles published between 2018 2023. The search procedure 124 relevant studies employed for diverse tasks including lesion detection, skin classification, retinal abnormality identification, cardiac brain tumor segmentation, personal protective equipment detection. findings demonstrated effectiveness outperforming alternative existing methods these However, review also unveiled certain limitations, well-balanced annotated datasets, high computational demands. To conclude, highlights identified research gaps proposes future directions leveraging potential

Language: Английский

---

A Comprehensive Survey of Convolutions in Deep Learning: Applications, Challenges, and Future Trends

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 41180 - 41218

Published: Jan. 1, 2024

In today's digital age, Convolutional Neural Networks (CNNs), a subset of Deep Learning (DL), are widely used for various computer vision tasks such as image classification, object detection, and segmentation. There numerous types CNNs designed to meet specific needs requirements, including 1D, 2D, 3D CNNs, well dilated, grouped, attention, depthwise convolutions, NAS, among others. Each type CNN has its unique structure characteristics, making it suitable tasks. It's crucial gain thorough understanding perform comparative analysis these different understand their strengths weaknesses. Furthermore, studying the performance, limitations, practical applications each can aid in development new improved architectures future. We also dive into platforms frameworks that researchers utilize research or from perspectives. Additionally, we explore main fields like 6D vision, generative models, meta-learning. This survey paper provides comprehensive examination comparison architectures, highlighting architectural differences emphasizing respective advantages, disadvantages, applications, challenges, future trends.

Language: Английский

---

UAV T-YOLO-Rice: An Enhanced Tiny Yolo Networks for Rice Leaves Diseases Detection in Paddy Agronomy

IEEE Transactions on Network Science and Engineering, Journal Year: 2024, Volume and Issue: 11(6), P. 5201 - 5216

Published: Jan. 10, 2024

The paddy agronomy in the Asia-pacific region has gained a prominent role connection with major rice production area over decades. research aims to investigate aerial computing techniques improve sky farming techniques. Recently, enhancement of unmanned vehicle (UAV) and Internet Things (IoT) Deep Learning (DL) ensured impact on data availability predictive analytics. In this research, we focus for identifying weeds, regions crop failure, health crops. Therefore, DL architecture suitable application UAV onboard intelligence is necessary. Furthermore, should be stable consume as few computational resources possible, given that it applied UAV's system. This paper proposes use Tiny YOLO (T-Yolo)V4 base detector via following modules: (a) detection layer added T-YOLO v4 make network more capable detecting small objects. (b) Spatial pyramid pooling (SPP), convolutional block attention module (CBAM), Sand Clock Feature Extraction Module (SCFEM), Ghost modules, layers are increase accuracy network. Subsequently, leaf diseases set which contains labeled images such Bacterial blight, Rice blast, brown spot obtained. addition, image augmentations produce samples three classes create our own set. Finally, enhanced Yolo (UAV T-yolo-Rice) obtained testing mean average precision (mAP) $86 \%$ by training proposed leaves' disease More experimental results reveal method outperforms Leaves' Diseases model using T-yolo-Rice can obtain highest Mean Average Precision than all other models from previous studies. Even V7 produced darknet cannot have higher

Language: Английский

---

Intrarow Uncut Weed Detection Using You-Only-Look-Once Instance Segmentation for Orchard Plantations

Sensors, Journal Year: 2024, Volume and Issue: 24(3), P. 893 - 893

Published: Jan. 30, 2024

Mechanical weed management is a drudging task that requires manpower and has risks when conducted within rows of orchards. However, intrarow weeding must still be by manual labor due to the restricted movements riding mowers orchards their confined structures with nets poles. autonomous robotic weeders face challenges identifying uncut weeds obstruction Global Navigation Satellite System (GNSS) signals caused poles tree canopies. A properly designed intelligent vision system would have potential achieve desired outcome utilizing an weeder perform operations in sections. Therefore, objective this study develop module using custom-trained dataset on YOLO instance segmentation algorithms support recognizing obstacles (i.e., fruit trunks, fixed poles) rows. The training was acquired from pear orchard located at Tsukuba Plant Innovation Research Center (T-PIRC) University Tsukuba, Japan. In total, 5000 images were preprocessed labeled for testing models. Four versions edge-device-dedicated utilized research—YOLOv5n-seg, YOLOv5s-seg, YOLOv8n-seg, YOLOv8s-seg—for real-time application weeder. comparison evaluate all models terms detection accuracy, model complexity, inference speed. smaller YOLOv5-based YOLOv8-based found more efficient than larger models, YOLOv8n-seg selected as evaluation process, had better accuracy YOLOv5n-seg, while latter fastest time. performance also acceptable it deployed resource-constrained device appropriate weeders. results indicated proposed deep learning-based speed can used object recognition via edge devices operation during

Language: Английский

---

Enhanced YOLOv8-based method for space debris detection using cross-scale feature fusion

Deleted Journal, Journal Year: 2025, Volume and Issue: 7(2)

Published: Jan. 21, 2025

Language: Английский

---

Integrating text parsing and object detection for automated monitoring of finishing works in construction projects

Automation in Construction, Journal Year: 2025, Volume and Issue: 174, P. 106139 - 106139

Published: March 23, 2025

Language: Английский

---

Evaluating YOLOV5, YOLOV6, YOLOV7, and YOLOV8 in Underwater Environment: Is There Real Improvement?

2022 7th International Conference on Smart and Sustainable Technologies (SpliTech), Journal Year: 2023, Volume and Issue: unknown

Published: June 20, 2023

This paper compares several new implementations of the YOLO (You Only Look Once) object detection algorithms in harsh underwater environments. Using a dataset collected by remotely operated vehicle (ROV), we evaluated performance YOLOv5, YOLOv6, YOLOv7, and YOLOv8 detecting objects challenging conditions. We aimed to determine whether newer versions are superior older ones how much, terms performance, for our pipeline dataset. According findings, YOLOv5 achieved highest mean Average Precision (mAP) score, followed YOLOv7 YOLOv6. When examining precision-recall curves, displayed precision recall values, respectively. Our comparison obtained results those previous work using YOLOv4 demonstrates that each version detectors provides significant improvement.

Language: Английский

---

An Ensemble Approach for Robust Automated Crack Detection and Segmentation in Concrete Structures

Sensors, Journal Year: 2024, Volume and Issue: 24(1), P. 257 - 257

Published: Jan. 1, 2024

To prevent potential instability the early detection of cracks is imperative due to prevalent use concrete in critical infrastructure. Automated techniques leveraging artificial intelligence, machine learning, and deep learning as traditional manual inspection methods are time-consuming. The existing automated crack algorithms, despite recent advancements, face challenges robustness, particularly precise amidst complex backgrounds visual distractions, while also maintaining low inference times. Therefore, this paper introduces a novel ensemble mechanism based on multiple quantized You Only Look Once version 8 (YOLOv8) models for segmentation structures. proposed model tested different datasets yielding enhanced results with at least 89.62% precision intersection over union score 0.88. Moreover, time per image reduced 27 milliseconds which 5% improvement other comparison. This achieved by amalgamating predictions trained calculate final mask. noteworthy contributions work encompass creation time, an robust segmentation, enhancement capabilities models. fast renders it appropriate real-time applications, effectively tackling infrastructure maintenance safety.

Language: Английский

---

Automated detection of railway defective fasteners based on YOLOv8-FAM and synthetic data using style transfer

Automation in Construction, Journal Year: 2024, Volume and Issue: 162, P. 105363 - 105363

Published: March 11, 2024

Language: Английский

---