.png){kind=icon}
.png){kind=icon}
Return
DiffKPD: A Robust Key Point Detection Algorithm With a Diffusion Process for Automatic Pin Welding
.png){kind=icon}
PRE
AI
DOI：10.1109/JSEN.2025.3531794.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Recent advancements in vision-guided automatic pin welding methods have garnered significant attention for their high productivity and performance. However, these applications still face considerable challenges in achieving the robustness and precision required for pin recognition and key point localization, which impedes their adoption in intelligent automation and manufacturing, particularly for electric vehicle motor production. In this article, we introduce a novel automatic pin key point detection model, namely, DiffKPD, designed to overcome these challenges. Our solution uses a two-stage detector: a base detector for localizing pin key points, followed by a lightweight, fast diffusion process that leverages time steps and local spatial context information to refine prior key point detections. Finally, we demonstrate the effectiveness and efficiency of our proposed method through extensive experimental results in terms of both localization precision and inference speed.
Keywords:
Automatic pin welding
diffusion process
key point detection
vision-assisted welding
Journal
IF:
4.5
Papers:
2.1W
Citations:
7.3W