Restricted Research - Award List, Note/Discussion Page

Fiscal Year: 2021

193  University of North Texas  (84489)

Principal Investigator: Xu,Bugao

Total Amount of Contract, Award, or Gift (Annual before 2011): $ 60,617

Exceeds $250,000 (Is it flagged?): No

Start and End Dates: - 8/31/21

Restricted Research: YES

Academic Discipline: Merch & Digital Retailing

Department, Center, School, or Institute: Col of Merch, Hosp & Tour

Title of Contract, Award, or Gift: On-Loom Fabric Defect Detection by CIS and Machine Learning

Name of Granting or Contracting Agency/Entity: Milliken & Company

CFDA:

Program Title: N/A

Note:

This project is Restricted Research. This project is Research- Applied. This project is to fine-tune the technology and to build three on-loom CIS scanners (LoomScan) for further testing in Milliken's Gerrish plant. The specific aims are to: * Design the control board that can connect up to 10 new CISs (Canon JC06H) which can cover a width of <=3000 mm. Since the new CIS is incompatible with the one we used, the control board has to be redesigned and tested out for multiple (10) CIS units. * Develop the software driver that can regulate the scanning rate, resolution, image transfer of 10 CISs simultaneously. * Design the framework of real-time communications of big data (600 dip images) between the CISs and the embedded computer. * Retrain the deep-learning model with the images captured by the new CIS system to detect kinky fillings, folds, and oily marks. It is expected to get a detection accuracy better than 96%. * Send the signal for a detected defect through the light pole to set off an alarm.

Discussion: No discussion notes

 

Close Window