Through-Hole Soldering Defect Detection (Bridge / No Solder) in LabVIEW – Vision or Python Integration?

[sasimitha]

Hi everyone,

I’m working on an automated through-hole soldering defect detection system using LabVIEW and I’d like your suggestions or guidance on the best approach.

The goal is to detect two main types of defects:

• Bridge defects – where solder connects adjacent pins unintentionally.

• No solder defects – where solder is missing or insufficient on a through-hole.

🛠️ Setup & Requirements:

• I have a dataset of annotated images (color and grayscale) showing good and defective solder joints.

• I’m using LabVIEW 2020 with the Vision Development Module (VDM).

• Images are captured from a top-down camera post-soldering.

🔍 Questions:

1. What would be the best approach using LabVIEW Vision tools (e.g., blob analysis, pattern matching, edge detection) to detect these solder defects?

2. Are there example VIs or recommended vision functions for through-hole solder joint analysis?

3. Has anyone implemented this using LabVIEW Machine Vision tools alone, or is an external model preferable?

4. If I train a deep learning model in Python (e.g., YOLOv5 or TensorFlow) for defect detection, how can I integrate the inference back into LabVIEW efficiently?

   • For example: using LabVIEW Python Node, REST API, or shared memory.

5. Are there any considerations for real-time performance or image preprocessing specific to through-hole joints?

🧠 I’m open to both traditional vision approaches and ML-based solutions, so any example VIs, articles, or integration tips would be greatly appreciated!

📸 Attached Sample: I've uploaded a sample image that represents the type of images we capture in production. The image includes a cluster of solder joints (not a single pin) which need to be individually analyzed.

[Jay14159265]

+1 for writing it all in python.