The textile is not only textile anymore due to the improved technology nowadays. Since green-tech, medical resource, and electronics components are added, a creative and functional textile is born. However, an astonishing textile catches people’s eyes by its quality, the defect-detection of textile is one of the most important things during the process of it; the defect-detection now work manually, which take times and easy to miss any blemish. So, Taiwan Textile Research Institute Department of Processing Technology Development (TTRI) pay efforts to develop smart textile inspection, they design an algorithm that can automatically detect the textile based on the big data they collected for years.
Taiwan took an important role amount the textile market nationwide, having unlimited creative power. In order to upgrade as industry 4.0, there is no time to waste to wait; therefore, TTRI plans to deploy a small size of an intelligent platform on the traditional textile inspection machine, which reduces the time of development period in the meanwhile to keep the deployment of the product line as usual.
Moreover, there are various kinds of textile on the market, the hardware of smart platform should equip with high compatibility and flexibly support different algorithms. The smart defect-inspection work with the machine vision, and due to the width of standard textile, TTRI needs the platform to drive five cameras to detect the images with a powerful efficiency inference.
Considering the various textiles characteristics, the limits of the deployment of the product line, and the multi I/O requirements, TTRI turned to Aetina's Jetson platform solution to drive their algorithm.
Aetina's Jetson platform supports up to six cameras with the EVID (extension video image detection) service, which fulfill the width of the standard textile. Various I/O connectors from the carrier board help TTRI to overcome the difficulties from the traditional inspection machine, successfully upgrade it to an intelligent auto-defect-detect machine under a lack time of development period and the limitation of machine specification.
A powerful GPU computing from the Jetson AI platform fasten the inference of textile image detection, not only detect the textile automatically but also collect the data from the process, bring an opportunity to manage the quality of textile production. Aetina’s carrier board support diverse I/O that meets TTRI’s requirements, fully liberate their algorithm design, in the meanwhile push on the industry 4.0 for all textile factory.