Exploring the Frontier of Object Detection: A Deep Dive into YOLOv8 and the COCO Dataset
Authors:
- 10 December 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Abstract
The field of computer vision has recently widened to incorporate object identification, which has significant impacts in areas such as autonomous vehicles, robotics, security, and healthcare. This research review and analysis look at how well YOLOv8 (You Only Look Once), the latest version of the popular object recognition model, does on the COCO data set. The purpose of this research is to learn about the YOLOv8 architecture, its improvements over previous versions, the COCO data set's make-up and evaluation metrics, and their strengths and weaknesses. From its first version through YOLOv8, the paper discusses the YOLO architecture's core features and enhancements. To establish a benchmark, the YOLOv8 model is compared to other top-tier object detection models as Faster R-CNN, SSD, and EfficientDet. The COCO data set is investigated in detail, including an overview of its pictures, annotations, item categories, data collection, and annotation procedure, and the evaluation criteria used to test object identification algorithms on this data set. The study also discusses YOLOv8 architecture and performance limits and COCO data set biases, data distribution, and annotation quality. Discussing the difficulties of generalizing YOLOv8 for diverse object detection tasks. YOLOv8 and the COCO data set are useful in real-world applications and case studies. The study concludes by offering possible enhancements for the YOLOv8 architecture, ideas for extending the COCO data set and assessment metrics and analysing upcoming trends in object identification research and their implications for YOLOv8 and COCO. This comprehensive review analysis helps object detection researchers, practitioners, and industry professionals choose models, datasets, and evaluation methods for their applications.Keywords
This publication has 11 references indexed in Scilit:
- A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NASMachine Learning and Knowledge Extraction, 2023
- YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors2023
- EfficientDet: Scalable and Efficient Object Detection2020
- YOLO9000: Better, Faster, Stronger2017
- Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors2017
- SSD: Single Shot MultiBox Detector2016
- You Only Look Once: Unified, Real-Time Object Detection2016
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Microsoft COCO: Common Objects in Context2014
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998