Deep Learning Algorithms With an Application in Garments Quality Control

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F46747885%3A24620%2F17%3A00004458" target="_blank" >RIV/46747885:24620/17:00004458 - isvavai.cz</a>

Výsledek na webu

<a href="http://www.iecesaudi.com/all-papers.pdf" target="_blank" >http://www.iecesaudi.com/all-papers.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Deep Learning Algorithms With an Application in Garments Quality Control

Popis výsledku v původním jazyce

Deep learning is a machine learning technique that utilizes many layers of non-linear transformations to extract features (in supervised or unsupervised manners) from the system’s input. This creates systems that process information more efficiently and capable of performing a wider range of operations for classification and pattern analysis purposes. The hierarchy in this technique with many (deep) layers sets its performance apart from the traditional machine learning techniques, such as the Artificial Neural Networks (ANN) that have "shallow architectures" based on one or two non-linear transformations. This work presents a case study for applying this technique, for the first time, in monitoring the quality control of garments and detecting their sewing defects. The introduced Artificial Intelligent (AI) system is based on reading the sewing line using a digital camera and processing the acquired images using the deep-learning algorithms. The system shows a great ability to transfer knowledge from pre-trained deep-networks to extract multiple features from the images and use these features in a successful classification of the sewing lines and highlighting the defected spots, if any. Results of this work opens the door for on-line detection systems that can work with higher efficiency, which should reduce the costs associated with salvaging defected garment products.

Název v anglickém jazyce

Deep Learning Algorithms With an Application in Garments Quality Control

Popis výsledku anglicky

Deep learning is a machine learning technique that utilizes many layers of non-linear transformations to extract features (in supervised or unsupervised manners) from the system’s input. This creates systems that process information more efficiently and capable of performing a wider range of operations for classification and pattern analysis purposes. The hierarchy in this technique with many (deep) layers sets its performance apart from the traditional machine learning techniques, such as the Artificial Neural Networks (ANN) that have "shallow architectures" based on one or two non-linear transformations. This work presents a case study for applying this technique, for the first time, in monitoring the quality control of garments and detecting their sewing defects. The introduced Artificial Intelligent (AI) system is based on reading the sewing line using a digital camera and processing the acquired images using the deep-learning algorithms. The system shows a great ability to transfer knowledge from pre-trained deep-networks to extract multiple features from the images and use these features in a successful classification of the sewing lines and highlighting the defected spots, if any. Results of this work opens the door for on-line detection systems that can work with higher efficiency, which should reduce the costs associated with salvaging defected garment products.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/LO1201" target="_blank" >LO1201: ROZVOJ ÚSTAVU PRO NANOMATERIÁLY, POKROČILÉ TECHNOLOGIE A INOVACE TECHNICKÉ UNIVERZITY V LIBERCI</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2017

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů