Benchmark Study on YOLOv8 Variants in Localized Multiclass Fault Detection in PCBs

Najib, Taeef; Muntasir, Fahim; Wasi, Wasif Al Wazed

doi:10.1109/iceeict62016.2024.10534448

Search citation statements

Order By: Relevance

Paper Sections

Select...

Citation Types

Supporting

Mentioning

Contrasting

Year Published

2024

Publication Types

Select...

Other1

Article1

Relationship

Self Cite0

Independent2

Authors

Journals

Cited by 2 publications

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Analysis of the various YOLOv8 models in Detecting Defects in PCB

Raj,

Ahlawat

2024

2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT)

View full text Add to dashboard Cite

Analysis of the various YOLOv8 models in Detecting Defects in PCB

Raj,

Ahlawat

2024

2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT)

View full text Add to dashboard Cite

Design of Deep Learning Techniques for PCBs Defect Detecting System based on YOLOv10

Ruengrote,

Kasetravetin,

Srisom

et al. 2024

Eng. Technol. Appl. Sci. Res.

View full text Add to dashboard Cite

As Printed Circuit Boards (PCBs) are critical components in electronic products, their quality inspection is crucial. This study focuses on quality inspection to detect PCB defects using deep learning techniques. Traditional widely used quality control methods are time-consuming, labor-intensive, and prone to human errors, making the manufacturing process inefficient. This study proposes a deep-learning approach using YOLOv10. Through the incorporation of architectural improvements such as CSPNet and PANet that improve feature extraction and fusion, as well as a dual assignments mechanism that increases localization accuracy, YOLOv10 offers significant improvements over earlier versions, such as YOLOv5 and YOLOv8, and Faster R-CNN models. These innovations allow YOLOv10 to deliver superior performance in terms of both speed and precision. The experiments used a custom dataset consisting of 1,260 PCB samples collected from the industry. The dataset was partitioned into 80% for model training and 20% for testing. The model was trained for 100 epochs with a batch size of 32 to evaluate its performance in identifying various PCB defects. YOLOv10, with its optimized architecture, fully utilized its capabilities while requiring less computational power than YOLOv5 and YOLOv8, especially in resource-constrained environments. Despite resource constraints, YOLOv10 achieved high accuracy, with a precision of at least 96% and a recall of 97%, surpassing earlier YOLO models and Faster R-CNN. It also achieved 99% mAP and more than 96% F1 score. These improvements in speed and accuracy make YOLOv10 a highly efficient solution for automated PCB inspection, reducing manual effort and offering fast and accurate classification adaptable to various applications.

show abstract

Benchmark Study on YOLOv8 Variants in Localized Multiclass Fault Detection in PCBs

Cited by 2 publications

References 17 publications

Analysis of the various YOLOv8 models in Detecting Defects in PCB

Analysis of the various YOLOv8 models in Detecting Defects in PCB

Design of Deep Learning Techniques for PCBs Defect Detecting System based on YOLOv10

Contact Info

Product

Resources

About