We embrace the challenge of bringing Industry 4.0 to the labour-intensive Asian apparel manufacturing industry. We are here to inspire and embark on the jounrey to advanced digitalization and building Smart Factories.
In the apparel industry, globally, executives inspect cent percent of the finished apparel manually by using measurement tapes.This consume a lot of time and is subject to human error. To make it time and cost effective, this project propose a system that uses machine vision-based decision making. The goal of this approach is to develop a real-time automatic garment measurement system based on artificial intelligence.
Research Team: Deepak Panghal & Prabir Jana
Fabric defect-inspection conventionally involves random checks conducted by executives on ground. This inspection requires experiential knowledge, is slow and only 70% accurate. Moreover humans are prone to auditory, cognitive and visual distraction, thereby hampering efficiency & repeatability.
The intelligent system captures the fabric defect through a camera, and analyse the defect using artificial intelligence.It can identify the defect with 95% accuracy. The system is faster, repeatable, can work round the clock and deskill the operation.
Research Team: Rajan, Ankit Prasad, Rashmi Thakur, Deepak Panghal & Prabir Jana
Sewing is the most time consuming and labour intensive process in apparel manufacturing. Real time monitoring of a sewing operator’s production is important for better utilisation of resources. Current processes uses barcode, RFID, NFC technology, where the operator’s action is necessary to actuate the scanning. IoT based sewing machines allpw the counting of production pieces by counting the multiplicity of backtacks.
The proposed system will map and count the production cycles from the machine speed data using artificial intelligence. The intelligent Sewing Production Monitoring System will be free from any operator error (advertent or inadvertent).It will increase accuracy by automatically filtering foreign elements and self-learning,
Research Team: Raghu, Deepak Panghal & Prabir Jana
Traditional apparel sewing is labour intensive.The sewing operator must manoeuvre the fabric through the needle point, increasing variance in stitches. The current scope is limited to flat garments made out of thermoplastic fibres.The proposed system conceptualises a model of fully automatic garment manufacturing system using welding technologies like ultrasonic and heat sealing. Using these systems, fabric rolls will directly pass through ultrasonic welding roller and garments will be produced at a very high rate.
Research Team: Vaibhav Agarwal, Deepak Panghal & Prabir Jana
Fabric residual shrinkage is a phenomenon that influences pattern making and garment measurement which has a considerable impact on consumer satisfaction. Fabric shrinkage is currently measured through ASTM D6207 test method that is destructive and time-consuming. This research is aimed at machine-learningbased prediction of fabric shrinkage through off-loom parameters like— fiber content, yarn count, weave, e.p.i., p.p.i. and gsm. This study extensively exploredmultiple ML platforms like IBM Watson, Jupyter Notebook, and MATLAB for prediction through supervised learning. Using the NN Tool of MATLAB The study found that the above-mentioned input parameters were unable to predict fabric shrinkage.
It is anticipated that adding larger data set with additional inputs of yarn crimp percentage, yarn twist and finishing details (sanforized or not, anti-shrinkage, crease-resistance, stain repellent, DMDHEU treatment) may enable better prediction.
Research Team: Akanksha Kumar, Rashmi Thakur, Deepak Panghal & Prabir Jana