PCB DeepScan

Speed up your QA process through our AI Powered Inspection tool using Cameras for Component Identification & Defect Detection

Why PCB DeepScan?

This is an AI Driven Automated QA Inspection solution for Component Identification and Defect Detection during the Printed Circuit Board (PCB) component assembly. With growing pressure on PCB suppliers to stay relevant & provide the best quality products, it leaves no scope for any imperfection or rework due to defects being neglected, which is quite possible during manual inspection due to the micro size of the components. Hence our tool enables defect detection at each stage of PCB batch processing avoiding post manufacturing rejection cost completely.

AI Powered

The solution enables PCB manufacturers speed up the QA process using our AI Powered Inspection tool that leverages cameras for component identification & defect detection. The tool is built using the following AI & Deep learning techniques:

  • ResNet
  • CornerNet
  • Object Detector
  • U – Net

Supplemented by Vision Analytics

The tool is also powered with the following Computer Vision techniques:

  • Thresholding & Morphological Operations
  • Template Matching
  • Background Subtraction
  • SSIM for Structural Similarities
  • Correlation Filters

100% Cloud Based

  • 24×7 global access and eliminating the concern of either software or hardware maintenance: Significant cost savings and greater flexibility
  • Scalability-on-demand through adding more Graphical Processing Units (GPUs) with 100% uptime

Functions of PCB DeepScan

Our AI visual inspection solution with the latest deep learning algorithm is capable of real-time inspection, identification, labelling and classification of PCB defects. The solution includes Cameras, lenses and an AI & Vision Analytics supported web portal. It identifies each component on the board and detects any defect during the assembling process itself.
Key Features of the Product:

  • Barcode Reading & PCB Identification
  • Component Segmentation & Classification
  • Defect Detection & Classification
  • Component Assembly Verification
  • Component Presence or Absence Detection
  • OCR for Component Information Capture

How PCB DeepScan works?

For PCB Manufacturers, defect detection is critical to ensure the finest quality products are delivered to their customer. PCB assembly is a complex process, and there are possibilities of defects occurring in the process, therefore rigorous inspection is paramount.

Implementation


1. Data collection

  • Images for Annotation, Training & Testing
  • CAD files for Component Segmentation, Component Identification and identification of Paste/Solder Areas
  • A golden PCB is used for registering the images of the PCBs from all stages of the assembly process & is used for anomaly detection

2. Corner Detection

  • ML processing to separate the Image to be inspected from the Background
  • Calculation of the PCB Corners

3. Manual Annotation

  • Manual annotation of Objects and Defects for Training the Deep Learning Model
  • Done using Annotation Tool

4. Synthetic Data Creation & Noise Generation

  • Scaling Object to standard size
  • Sampling to reduce image size for storage
  • Photometric noise synthesis: (shades, blurring, spots etc.)
  • Rotation & Alignment

5. Object Detection & Segmentation

  • Build Deep Learning Model for Object Detection & Identification
  • CAD files processing for capturing the Component boundaries and Paste/Solder Areas

6. Defect Detection & Classification

  • Model Training using Deep Learning & Vision Algorithms for Defect Detection & Classification

7. Deployment & Setup

  • Sophisticated Cameras & Lenses installed at different stages of the production flow to precisely monitor the PCBs & components for defects
  • Provisioning training to the PCB manufacturing floor in charges and QA team on the tool

To Know More – Download the Product Flyer