Client Support

+91-44-43858774

Contact Us

AI-Assisted Wiring Harness Revision Comparison for Automotive Engineering

Accelerating Engineering Review of Wiring Harness Revisions with Automated Change Detection 

Automatically identifying likely change areas between automotive wiring harness drawing revisions while providing engineers
with a measurable, ranked review aid.

Challenges

  • Manual Change Verification: Engineers manually compare old and new harness drawing revisions, highlight every change, and prepare change summaries. This makes the review process time-consuming and increases the risk of missed changes or documentation errors. 
  • Redrawn Revision Geometry: Direct drawing comparisons become unreliable when revisions are redrawn rather than edited in place. In the evaluated sample, drawing content shifted by approximately 50 pixels across the sheet, while an entire general-notes block had moved, creating significant comparison noise.
  • Low-Resolution CAD Plots: The source drawings were supplied as 200 dpi bilevel raster plots. Disconnected vector paths, raster artifacts, compressed CAD fonts, and limited image quality reduced the accuracy of both image-based comparison and text recognition.
  • High False-Alarm Rates: Conventional image comparison flagged 137 of 232 drawing zones as changed despite only 28 representing genuine revision updates, creating excessive review noise and making it difficult for engineers to focus on the areas most likely to contain real changes. 
  • Inconsistent Text Recognition: OCR extracted 23 part numbers from one revision and 19 from the other, but only 9 matched identically across both drawings. This limited the reliability of text-based methods for identifying and validating revision changes.

The Solution

iTech evaluated multiple approaches against the customer’s own manually marked drawing rather than relying on subjective assessment. The marked-up sheet contained 39 highlighter marks across 31 change zones, creating a customer-defined benchmark for measuring automated review accuracy. The proof of concept was designed as an engineering review aid that prioritizes likely revision locations; it does not automatically determine or validate the exact engineering changes.
  • Benchmark-Driven Evaluation: Traditional image comparison, two independent change detectors, and OCR-based text recognition were measured against the customer's marked drawing to assess detection accuracy and false alarms.
  • Dual-Detector Review Aid: The two-detector approach identified 26 of 31 customer-marked change zones. Areas where both detectors agreed were ranked higher to help engineers focus on the most likely revision locations.
  • Structured Review Outputs: The revised drawing was delivered as a highlighted PDF with candidate change areas ranked by detector confidence. A companion Excel workbook provided zone-level details including rank, detector agreement, changed-content measurements, and text unique to the current revision.
  • Measurable and Transparent Results: A dedicated Read-Me tab documented the benchmark methodology, detection performance, false-alarm rates, and known limitations, allowing stakeholders to assess the effectiveness and boundaries of the approach objectively.

Business Outcomes & Impact

  • 84% Changed-Zone Detection: Benchmark testing showed the dual-detector approach identified 26 of the customer's 31 marked change zones, achieving an 84% changed-zone detection rate.
  • More Focused Engineering Reviews: Ranked candidate locations gave engineers a prioritized starting point for reviewing revision changes instead of requiring an unguided scan of the entire drawing.
  • Reduced Review Noise Compared to Simple Image Comparison: Benchmark testing demonstrated that conventional image comparison generated excessive false positives, while the dual-detector approach provided a more practical method for narrowing the engineering search area.
  • Structured Review Workflow: The combination of highlighted drawings and detailed Excel outputs supported a more organized review process, giving engineers both a visual reference and supporting data to evaluate potential changes efficiently.
  • Clear Roadmap for Further Automation: The evaluation demonstrated that achieving higher-confidence automated change identification will require comparison of structured engineering data, such as native CAD or vector-based drawing information, rather than relying solely on raster image comparison and OCR.
iTech India
Powered by  GDPR Cookie Compliance
Privacy Overview

iTech uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognizing you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.