The study on surface defect detection of stamped parts based on improved deep learning

Robotic Process Automation for Smarter Devices in Manufacturing

Open Access

Table 5

Blur Type Test.

Blur type	Model	Precision	Recall	mAP50	mAP50-95
No Blur	YOLOv8 + FFA + Gold-YOLO	0.90	0.85	0.88	0.50
	Baseline model	0.82	0.75	0.80	0.45
Gaussian Blur	YOLOv8 + FFA + Gold-YOLO	0.88	0.83	0.85	0.48
	Baseline model	0.79	0.72	0.76	0.43
Motion Blur	YOLOv8 + FFA + Gold-YOLO	0.87	0.80	0.83	0.47
	Baseline model	0.74	0.68	0.73	0.41

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.