Application of AI in Quality Control in Textile & Apparel Industry

Application of AI in Quality Control in Textile & Apparel Industry

Ms. Sayoni Nath & Dr. Anirban Dutta
Government College of Engineering and Textile Technology, Serampore, W.B., India


The application of different elements of Industry 4.0 has a significant impact in the rapid technological upgradation in apparel and textile industry. It is reported that the worldwide textile market is projected to increase from $ 1.5 trillion in 2025.

Also, an article by Grand View Research indicates that, as a consequence of the increasing clothing demand in developing countries such as China, India, Mexico, and Bangladesh, the global textiles markets were anticipated to reach USD 961.5 billion in 2019. And it is projected to show a CAGR (compound Annual Growth Rate) of 4.3% from 2020 to 2027.

In this context, the gradually increasing demand for quality products has resulted in widespread application of automation and Artificial Intelligence (AI) in the apparel and textile industry. The prime objective of the application of Industry 4.0 elements in apparel and textile industry is to reduce the labour and manufacturing costs and to ensure quality standards of the finished products as per the standards based upon customers’ demands. The rise of new technologies, like AI and the Internet of Things (IoT), has resulted in a paradigm shift in the textile business. Which once was labour-intensive and highly depended upon human effort only.

In the present day, most textile and garment industries are equipped with computerized machines with Computer Aided Design (CAD) features. Also, the application of new technological tools like AI, IOT, Machine Learning (ML). Augmented Reality (AR) etc. has resulted in efficient production monitoring and increase in production efficiency in turn through real-time data capture, preventive maintenance, process optimization. Intelligent quality inspection system etc. In addition to that, certain applications of AI in optimization of product costs. fabrication of textiles, quality control. Just-in-time production in case of lean manufacturing, and CIM (Computer Integrated manufacture). Other unique and special AI application includes fault detection, pattern checking, and color matching, quality control and quality assurance in case of both fabric and garment. Artificial intelligence (AI) is gaining gradually increasing popularity. over the last two decades, in the textile industry. In many instances of textiles production, there are huge chances of error. The application of the AI can deal with the production process without error. As a result, over the last decade the use of AI is rapidly growing in textile industries for various applications.

In this present review paper, the different aspects and domains of the application of Artificial Intelligence in the quality inspection and quality control of different apparel and textile products have been discussed based upon available research and articles in this domain.

Introduction to Artificial intelligence (AI):
Just like the enormous complexity and delicate precisions in case of human intelligence, it must be noted that AI is also a combination of several algorithms of problem-solving, decision making, and reasoning which is computer-controlled as well as robot-based. AI (Artificial Intelligence) refers to the ability of a robot-based and computer- based technology to perform certain tasks, take logical decisions and gene- rate results that are usually done by intelligent beings. This is mainly used to generate decisions based on a complex understanding of several processes and input parameters culminated.

Artificial Intelligence refers to a miniature simulation of human intelligence in smart machines which are programmed to analyse, recognise and respond like humans, and replicate their actions. It is a wide-ranging branch of computer science which concerned with building machines capable of performing tasks that typically require human intelligence. Also, AI enabled systems are capable of learning from experience and show logical outcomes or prediction.

Application of AI in Quality Control of Textile & Apparel Products:
The application and potential of AI in quality control of different textile products is very significant at the modern textile and apparel industry. The present section of this paper has presented a product-wise application of AI, based upon available research articles.

AI in yarn quality control:
During each phase of the yarn manufacturing, the production process has been entirely revolutionized by the use of AI. In each of the sequential stages, like blowroom, carding, drawing, combing, speed framework, ring spinning, winding, packing, and conditioning, the required production parameters are determined by AI-Based control panels, with little or no human participation. The production parameters based upon the required yarn quality and the raw material parameters can be determined and standardised by the AI enabled systems in yarn manufacturing process. The selection of optimum process parameters results in better quality of resultant yarn.

Also, the prediction of yarn quality by computing spinning parameters has become more accurate. The use of artificial intelligence has reduced yarn grading mistakes to as much as 60%, leading to better textile grading and more efficient selection of yarn for further process.

Also, the recent advancements include the use of optical microscopes. Fourier transforms infrared, and Raman spectroscopy for fibre inspection and fibre grade identification. Furthermore, Al can also be used to identify and grade textile fibres according to their color and other properties such as fineness, length, uniformity ratio, tenacity, and effect of spinning performance on yarn properties. The proper selection of the raw material, i.e., fibre is the first and most significant step towards production of high-quality yarn.

Apart from the selection of process parameter and selection of fibre, there have been several applications of Al in yarn manufacturing that includes virtual modelling or simulation of yarn from fibre properties, prediction of yarn tensile properties, prediction of yarn unevenness, and yarn engineering. This also helps a lot in pre-assessment of yarn quality, so that it is possible to alter the process parameters in the planning stage only to achieve target yarn quality.

A specific application of Al in case of dynamic yarn quality control has been discussed by Mr Tesfay Welamo”. As the sliver evenness is a very important parameter so far as the resultant yarn quality is concerned. Therefore, the Al enabled auto-levellers in the modern carding and drawframe machines can be very effective tool. Auto-levellers mounted on modern carding and drawing frames can be accurately adjusted and auto-controlled by Al enabled computerised systems to help to ensure yarn uniformity. The distance between the scanning rollers pair and the point of draft is called the Levelling Action Point (LAP). The Levelling Action Point (LAP) is one of the most crucial auto-levelling parameters which significantly controls the evenness of the slivers produced. In this context, Artificial Neural Networks are applied to predict the optimum value of the Levelling Action Point for different production lines, material conditions and required yarn parameters.

Also, the high performance computerised system aided by artificial intelligence is gaining popularity for the analysis of fibre properties and structure. This Al enabled system is capable of effective analysis of the fibre fineness analysis, identification of blend composition in a fibre blend, fibre identification and simulation of the yarn structure and probable defects in yarn manufacturing.

SS Doke and N Sanmugam highlighted the potential of Al tools like Expert System and ANN for the yarn and fabric quality control. Some of the commercially available expert systems developed by Uster are Quantum Expert. Lab Expert. Sliver Expert and Ring Expert systems. Quantum Expert system is capable of integration of the input Information received from rotor spinning machines and winding machines, and subsequently can generate the prediction model for machine performances depending upon the raw material qualities. Thus, corrective actions can be planned even before the start of the production in order to avoid quality deterioration. Moreover, the Expert system of Uster can be used to select the optimized yarn clearer setting in winding machine as per the required yarn quality parameters. Lab Expert system of Uster is used for simulation and pre-assessment of yarn and fabric qualities based upon several statistical data collected from different production machines. In this context, the sliver expert system of Uster is used for controlling the sliver count, short thick faults and the monitoring of productivity, production time loss due to machine stoppages and Identification of probable causes of the faults in both process and product. The algorithm of sliver expert system is based upon the information collected from the carding, comber and drawframe machines. In this sequence. the ring expert system of Uster is used to detect the idle spindle, slipping spindle and pre-identification of the excessive end-breakage.

On the other hand, as proposed by S S. Doke and N Sanmagam, the ANN tools can be effectively used for ensuring quality standards of textile products. Especially in the domains of cotton fiber grading, prediction of yarn CSP, yarn grading, fabric color-fastness grading, fabric comfort prediction and fabric inspection, ANN tool can be successfully applied to generate prediction models for the prediction of objective grading of cotton fiber based upon the input parameters like staple length, fineness, strength, trash percentage etc. In case of the prediction of CRP, it is reported that the research conducted by CIRCOT has revealed that the CSP prediction model developed using ANN tool can reduce the prediction error by 60% compared to other statistical regression models. Similarly, the report conveys that a very accurate yarn grading prediction model can be developed using ANN too based upon certain measurable parameters like imperfection, U%, hairiness value, classimate values etc.

AI in inspection of fabric:
An inferior quality fabric can result in substandard garments as well as reduces production rate during garment manufacturing. Any undetected defect in the fabric is passed into the final garment stage, can result in the rejection of the finished garment. Hence, it is essential to build an error free system for detection of fabric defects at the fabric inspection stage before issuing the fabric to the spreading and cutting process for garment manufacturing.

Generally fabric inspection is performed by skilled workers using lighted tables or manually driven fabric inspection machine. This process is rather slow and not free from human error. Hence, the process is vulnerable and many times it can allow faults to pass to the garment manufacturing section. Furthermore, the rate of fabric inspection is directly influenced by the human fatigue in case of manual fabric inspection. In this context, the application of Al can perform this task at a quicker rate with much higher accuracy, and without fatigue. Fig 1 represents different types of fabric defects detected by the Al enabled automated fabric inspection system.

Different fabric defects inspected and identified artificial intelligence: (a) gout, (b) warp float, (c) draw back, (d) hole, (e) dropped stitches, and (f) press-off
Figure 1: Different fabric defects inspected and identified artificial intelligence: (a) gout, (b) warp float, (c) draw back, (d) hole, (e) dropped stitches, and (f) press-off

An important raw material for the clothing industry is fabric. The quality of the fabric influences the quality of the garment, productivity and the ease with which garments can be manufactured.

The fabrics are selected based on the type of garment and their end-use applications. The fabric specifications for making any garment can be classified as primary and secondary. The physical dimensions are considered to be primary. Whereas the fabric reaction to external forces is considered to be secondary. From a consumer perspective, garment appearance, comfort, and durability are important parameters. AI can be applied to control these parameters:

Fabric is the most important textile material. As it is the final product of textile manufacturing, and also it is prime raw material for apparel manufacturing. There it is of ultimate importance to ensure flawless fabric inspection system. In case of manual inspection of textile fabrics, the lack of concentration, human fatigue, human error, lack of uniform standards for passing or rejecting a fabric and quite a high time consumption are the main drawbacks observed. In this context, as a result of continuous research during the last two decades, various computer vision-based applications are proposed in various research articles to overcome the limitations and drawbacks associated with manual fabric inspection systems. The proposed study by Mr Tesfay Welamo” presents a detailed overview of histogram-based approaches, colour-based approaches, image segmentation based approaches. frequency domain operations, texture-based defect detection, sparse feature-based operation, image morphology operations, and recent trends of deep learning. The system architecture of a typical automated textile-fabric inspection system is shown in Fig 2.

As shown in Fig 1 the system consists of a series bank of cameras arranged in parallel across the web to be inspected, a computer console hosting (single or an array of) processors, the frame grabber, a lighting system and the supporting electrical and mechanical interfaces for the inspection machine. The inspection system employs massive parallelism in image acquisition with a front-end algorithm that reduces and limits the data flow to the region of interest only. The key components of this system are briefly reviewed by the author, as it is summarized below.

Architecture of Al enabled fabric inspection system
Figure 2: Architecture of Al enabled fabric inspection system

Lighting system: The quality and clarity of acquired images plays a vital role in AI enabled image processing- based inspection system. In this domain. the quality of illumination largely affects the acquired image quality. There are four common types of lighting schemes used for visual inspection i.e. front, back. fibre-optic, and structured. The back lighting eliminates the shadow and glare effects, and is widely used for fabric inspection. It is also possible to employ fibre optic illumination for the fabric inspection even though it is comparatively expensive, as it provides uniform illumination of the object under inspection. Also, it eliminates shadow or glare problem. A fuzzy logic control scheme using a feedback photo-resistor is sometimes used by the illumination controller to maintain a uniform (within 1% of deviation margin) level of illumination.

Camera: The resolution of a camera is a very important factor for any Al enabled fabric inspection system. The resolution is limited by the number of pixels in the camera photosensor and the object Field of View (FOV). The FOV is dependent on the characteristics of the background and the nature of defects to be detected. There are two common types of scanning techniques adopted for the cameras in case of computerised fabric inspection system. i.e. line scanning and area scanning. In case of the line scanning techniques which utilize a system of linear array photosensors, the resolution in the vertical direction is a function of the velocity of the fabric movement and the scan rate (line rate) i.e. the rate at which the camera is operating. Even though the modern line scan cameras usually provide very high resolution and can inspect a large portion of textile web in the single line scan, there is certain disadvantage as well associated with line scanning system. The major disadvantage with the line scan cameras is that they do not generate complete image at once and needs external hardware to build up images from multiple line scans. Also, line scanning system needs transport encoders for proper synchronization between camera speed and the speed of the fabric. On the other hand, in case of the area scan cameras, the usage of transport encoders is optional and the inspection resolution in both directions is independent of object (web) speed. Furthermore, the area scan system is more economical and user-friendly compared to line scanning system. Both CCD or CMOS photosensors can be used successfully. CMOS active pixel architecture provide higher level of on chip functionality at lower cost and low power usage than those from the CCD photosensors. However, the CMOS sensors are generally less sensitive than their CCD counterparts. Mainly due to higher uniformity and smaller fill factor. The inspection of fabric defects using CMOS area scan cameras, time-delay and integration (TDI) line scan cameras. have been attempted by the researchers. The typical system architecture of a computerized fabric-inspection system is represented in Fig 3.

Transport encoder: The transport encoder is used to provide master timing pulses for the camera. The wheel of the transport encoder is in direct contact with fabric winder. In case of line scan cameras, the resolution of the transport encoder (i.e., number of pulses per revolution) determines the pixel resolution. The line scan cameras can acquire crisp images at any speed by slaving camera scan rate to transport velocity. The velocity information from the transport encoder is also used to control any undesirable variation in the speed of shaft rollers.

The typical system architecture of computerized fabric inspection system by (a) manual and (b) automated optical inspection methods
Figure 3: The typical system architecture of computerized fabric inspection system by (a) manual and (b) automated optical inspection methods

Frame grabbers: The pixel data coming from each of the camera is converted into a digitized image by the frame grabber. All web inspection systems, such as the one used for fabric have to cope with the multiple camera inputs. Some systems do this by using some kind of video multiplexer unit between the camera and the frame grabber. A rather expensive way to cope with multiple cameras is to use one Trame grabber unit per camera. This permits parallel processing of image pixel data if the system is equipped with the multiple processors. The output from the frame grabber is Transported to the host computer in any of the popular PC formats ISA VESA, PUT etc. For industrial bus formats IN ME. PICMG, PC100, etc.

Host computer: The functions of the host computer can be classified into three main categories:

Defect detection and classification: The image data downloaded from the frame grabber is processed by the host computer, in order to perform defect detection using sophisticated algorithms. The detected defects are also categorized into different groups depending upon the parameters like origin and size

Camera illumination and control: The host computer is responsible for the external loading of the control setting parameters of the camera: either automatically or manually through Graphical User Interface (GUI). Moreover host computer is also responsible for the settings of the illumination controller, which controls the illumination level of web.

System control: The host computer is also responsible for a number of input and output system control functions The functions in this category include Interrupt Service Routine (ISR) Graphical User Interface (GU) and printing storage of the compressed defect map etc. As a single general purpose host computer is not sufficient to process large volume of image data acquired to inspect the textile fabric moving at the speed of 15-20 meters per minute. Most fabric-inspection systems use a single separate processor to detect all defects present in images from an individual camera. Each of these processors usually requires additional DSP processors (such as TMS320C40. AT&T 32C, etc.)/ for real time implementation of sophisticated defect detection algorithm”.

GM Nasira and P Banumathi proposed a computer aided inspection process for the fault detection of woven fabrics. The image processing technique and the Artificial Neural Network (ANN) algorithm have been used in this proposed inspection process. As per the process sequence proposed by the researchers, at first the high-quality vibration-free image of the fabric is acquired. Initially 300 dpi (Dot Per Inch) resolution is used by the researchers. Resolution is gradually increased step by step at the rate to 100 dpi per step, upto 1000 dpi. The different types of camera like CCD (Charged Coupled Device), CMOS (Complementary Metal Oxide Semiconductor), digital camera etc are used for image acquisition.

Next the acquired image is normalized and pre-processed by the image analysis technique, followed by conversion of pre-processed image into binary image using threshold values. The Bilinear interpolation method is used for normalization of the acquired images. The prime objective of this stage is to enhance the image features.

This method operates on the closest 2 x 2 neighborhood of known pixel values surrounding the unknown pixel. The final interpolated pixel value is determined by the software using the weighted average of the 4 pixel values. The Gaussian Filtering technique is used to reduce and suppress the noise from the processed image. The original digital image is transformed into grey scale image. The corresponding binary image is generated using the threshold value of the grey scale image, which is calculated by histogram method.

From the binary version the relevant features are extracted which includes the area of faulty portion, number of objects, shape factor, mean and variance are used as the input-set for training of the neural network model. These attributes are then used as the input parameters for the ANN model, which uses back propagation algorithm to determine the weighted factors to determine the outputs for the fault recognition in fabric.

Total 115 number of images of both defected and defect free fabric samples have been used by the researchers for training of the proposed ANN model. This research proposes a multilayer neural network, consisting of one input layer, two hidden layers and one output layer. The first and second hidden layer consists of 20 and 5 neurons respectively. The five neurons of the second hidden layer represent five different categories of probable outcomes of fabric inspection like hole, stain, warp float, warp float and no fault. The output layer generates the target output as binary string.

The proposed sequence of computerized fabric inspection is represented by Fig 4.

Proposed sequence of computerized fabric
Figure 4: Proposed sequence of computerized fabric

It is reported by the researchers that the proposed ANN model is capable of identifying the hole-type fault with 89% accuracy. Likewise, the success percentage for recognizing the stain, warp float and weft float are reported as 92%, 90% and 90% respectively, with an overall performance rate of 90%7.

Pattern inspection: Fabric pattern may have multiple aspects such like: weaving, knitting, braiding, finishing, and printing, etc. By replacing manual inspection with AI enabled image-based inspection could help manufacturers avoid human fatigue and errors in the detection of novelties and defects. AI techniques like Artificial Neural Network (ANN) are applied for defect identification through pattern recognition in fabric inspection of the textile industry. The fabric picture to be analysed is obtained from the image acquisition system and saved in relevant standard image format (jpeg, jpg, png etc). Different multi-Layer back propagation algorithm is used to train and test this ANN system. The system learns the weaving pattern, yarn properties, colours and tolerable imperfections from these images.

Prediction of fabric properties: AI can be used to predict the fabric properties before manufacturing with the help of neuro-fuzzy or other approaches by using the constructional parameters of the raw-materials i.e. fibre, yarn, and also using fabric constructional data. While applying AI, it is essential to establish a proper linear and nonlinear relationship between the relevant input parameters of fibre and yarn and the required fabric properties as per the end use. AI can also be applied to investigate comfort properties of the apparel fabrics. While sensorial comfort is considered, the fabric can be classified according to their hand value by the application of AI.

Application of AI in colour matching of textile products:
Colour is the first and one of the most significant elements of design, so far as the consumer response is concerned. Hence it is one of the important features of textile and apparel products. Consumers select or reject clothing or other fashion accessories on the basis of aesthetics which mostly includes color appeal. Hence, for getting the right color, precise quality control during dyeing and printing is essential, which can affect the buying decisions of the consumers and ultimately the volume of sales. Both the dyeing and printing process must ensure that the required color-fastness, depth of shade, color matching, and surface characteristics have been achieved. These parameters are influenced by the dye and fabric combinations and the chemical reactions and interactions governing them. Deviation of these parameters from the allowable limit may lead to reprocessing or rejection of the whole fabric batch³.

In this domain, the use of AI can play a very significant role. AI can be used for recipe prediction: process control during dyeing and printing: color matching; and evaluation of the final dyed or printed fabric. One of the applications of AI for a color solution is during the fibre blending stage, while the roving is converted to yarn. The use of Al can assist in predicting the color produced when fibres of different colors are mixed together. In the case of a homogenous blend. the prediction of color can be performed more accurately by using theoretical and empirical models.

Moreover, Al can be used for color matching of fabrics and shade sorting. It can be used for true color production by predicting the concentration of dyes from their spectrophotometric absorbance’.

Colour is an important aspect of textile products. The appearance of a textile product is perceived to be related to its quality. The colour of a product is judged to be acceptable/unsatisfactory. or it can be judged in more details to be: ‘too light ‘or, ‘too dark’, ‘too red’ or. ‘too green’. To solve this problem. Al can be developed that has ‘Pass/Fail’ feature to help improve the accuracy and efficiency.

AI in colour management:
Data color was widely utilized for color management to validate that the original colour design is consistent with the completed textile colour. Datacolor proposes to take into consideration its AI function the historical data from visual evaluation outcomes of human operators and to generate tolerances that in turn lead to contributive inspections that are closely matched to the visual inspection samples.

Colored dyers, liquid pigments, dough, and media are contained in the Data Color System dispenser. Prepare the most accurate solutions for the various areas in less time.

AI in garment final inspection:
The inspection of finished and semi finished textile product during their production is essential to get fewer rejections. The final quality inspection of finished garments is mainly done by experienced people, which is very time-consuming and may be influenced by the physical and mental condition of the inspector. As a result, automated Al inspection is essential to achieve the efficiency and accurate results. Automated inspection can be performed by the use of AI and image processing for inspection of the quality of the product.

The inspection of finished and semi-finished garments during their production is essential to get fewer rejections. The final quality of a finished garment depends on the sewing quality and other faults present in it. The final quality inspection of finished garments is mainly done by experienced people, which is very time-consuming and often subjective in nature. The results of the inspection are influenced by the physical and mental condition of the inspector. Therefore, automated inspection devices are essential to achieve increased efficiency and accurate results. Although limited studies have been done, an automated inspection can be performed by the use of AI and image processing for inspection of the quality of finished garments.

During garment production, each process (cutting, sewing, and pressing) plays a vital role influencing the quality of the finished garment. The quality of the semi-finished products should be inspected at each of these processes before the final inspection. The finished garments are inspected as per their specifications, overall appearance, faults, and sizing and fit. In detail the finished garments are inspected for stitching quality, mismatched plaids or stripes along the seam, puckered seam or extra material caught in seams, uneven seam along hems, and many other faults that can arise during the garment production.

The application of AI in final garment, inspection includes: automatic classification of general faults in shirt collars (for mono-colored materials) using machine vision; application of AATCC (American Association of Textile Chemists and Colorists) wrinkle rating for evaluation of wrinkle by using a laser sensor; detection and classification of stitching defects using wavelet transform and BP NN: seam pucker evaluation by using self-organizing mapping: and designing of a smart hanger for garment inspection. In manufacturing seamless garments, Al can be used to detect faults online. The image of the final garment can be captured and compared with the standard and any variation from the standard is reported as a fault that can be mended at that time or a marking is done where the fault occurs.

Performance of sewn seam: In case of garment manufacturing, the seams and stitches are used to assemble two or more pieces of fabric together. The ease of seam formation and the performance of the seam in terms of seam strength, seam puckering etc are the important parameters, which are collectively judged by the term known as ‘sewability’. The low-stress mechanical properties of fabric such as tensile, shear, bending and surface can affect sewability. Al system can be used to find the sewability of different fabrics during garment production. Fabric mechanical properties affect the performance during different stages of apparel manufacturing like spreading, cutting, and sewing.

A good quality seam is essential for a good quality garment. The performance of a sewn seam depends on the type of fabrics, fabric construction and sewing thread properties. Also, the seam and stitch type and sewing parameters, which include needle size, stitch density, and the sewing thread tension, machine RPM. The performance properties of the seam are evaluated by seam puckering, seam slippage, and yarn severance, which can be predicted by AI. Such Al enabled prediction models can be used to pre-assess the seam qualities and seam performance, which in turn can play an important role in the apparel quality control’.

Commercially available AI applications:
The present segment of the article includes some specific and commercial applications of AI in the quality inspection of textile and apparel products.

Cognex – Fabric Pattern Inspection: Cognex Corp., the American manufacturer of machine vision systems, software, and sensors, established in Boston in 1981 offers its customised machine vision-based Cognex ViDi platform designed for fabric pattern recognition in the textiles industry. As claimed by Cognex, the Cognex ViDi platform can automatically inspect different aspects of fabric patterns such as weaving, knitting, braiding, finishing, and printing. The company also claims that its platform requires no development period for integrating it into a manufacturing system, and it can be trained using predefined images of what a good fabric sample looks like.

Datacolor Al enabled tolerancing for fabric colormatching: Datacolor founded in Lucerne, Switzerland in 1970 offers color management instruments and software. To ensure that the original design colours match the colours in a finished textile product, the organizers usually assign a ‘colour tolerance’ – a limit to determine the acceptable threshold value of difference in colour between a sample and the required standard. The AI enabled system can decide accordingly regarding the colour matching.

While traditional color tolerancing was done based on numeric descriptions of color through instrumental tolerancing systems, that method generally had a lot of false positives compared with visual inspections, causing delays in the approval process because of the need for careful human intervention. Datacolor claims it has developed an artificial intelligence Pass/Fail (P/F) feature to help improve the accuracy and efficiency of instrumental tolerance.

Furthermore, Datacolor suggests that its Al feature can take into account historical data of visual inspection results from human operators while creating the tolerances that in turn results in instrumental inspections matching more closely the samples of visual inspections”.

Datacolor’s Al P/F procedure purportedly works as follows:

  • The textile expert first visually reviews all the individual batches that had been manufactured
  • The operators enter the color measurements and tolerances for all the batches in the Datacolor software to help train the AI P/F system
  • The AI P/P system can then be tested for new batches to automatically set Al tolerances, training the system to determine which samples pass and fail”.

AI enabled system to automate quality control process:
The Hong Kong Polytechnic University recently developed an intelligent fabric defect detection system, called ‘WiseEye’, which leverages advanced technologies including Artificial Intelligence and Deep Learning in the process of quality control in textile industry. The system effectively minimises the chance of producing substandard fabric by 90%, thus substantially reducing loss and wastage in the production. It helps to save manpower as well as enhance the automation management in the textile manufacturing

Supported by Al-based machine-vision technology, the novel ‘WiseEye” can be installed in a weaving machine to help fabric manufacturers to detect defects instantly in the production process. Through the automatic inspection system, the production line manager can easily detect the defects, thus helping them to identify the cause of the problems and fix them immediately”

Textile manufacturers currently rely on human efforts to randomly inspect the fabric by naked eyes. Due to human factors such as negligence or physical fatigue, defect detection by human labour is usually inconsistent and unrehable Textile manufacturers also attempted to use some other fabric inspection systems, but those systems were not able to meet the industry needs. Ensuring quality in the fabric production becomes a great challenge to the industry.

Prof Calvin Wong said. ‘Wise+Eye’ is a unique Al-based inspection system that satisfies the requirements of textile manufacturers. It is an integrated system with a number of components that perform different functions in the inspection process. The system is embedded with a high-power LED light bar and a high-resolution charge-coupled device camera which is driven by an electronic motor and is mounted on a rail to capture images of the whole width of woven fabric during the weaving process. The captured images are pre-processed and fed into the Al-based machine vision algorithm to detect fabric defects. Real time information gathered throughout. the detection process are sent to the computer system, and analytical statistics and alert can be generated and displayed as and when needed

The research team has applied Big Data and Deep Learning technologies in “Wiselye. By inputting data of thousands yards of fabrics into the system. the team has trained ‘Wiselye to detect about 40 common fabric defects with exceptionally high accuracy resolution of up to 0.1 mm pixel”

And finally the Fabrics can is a system developed by Uster using ANN algorithm for the fabric inspection process. In this process CCD camera is used to scan the fabric under inspection, followed by feeding the scanned signals as input set into the ANN network. The output layer of the ANN initiates marking of the fabrics by using defect marker as and when objectionable defect is identified by the ANN networks.

AI can be used in various segments of the quality control and quality inspection of textile and apparel products. Developed countries have already adopted AI enabled systems to improve the quality of garments, fabric and yarn. significant progresses are undergoing in Al rapidly and in near future it is expected that the AI will become an important tool for the garment manufacturers for quick, automated and accurate quality inspection and quality control.

AI system is one of the best choices in the textile and apparel industry to integrate the features like production. quality, cost, etc, to cope up with lean manufacturing, just-in-time production and computer integrated production. Hence it is needless to mention that the application of artificial intelligence in textile industry has a huge potential similar to other areas of application.

It seems clear that real-world AI applications in the textile sector are still at a very initial stage, and it is possible that cutting-edge AI manufacturing applications are more likely to be adopted by the textile and apparel industry for quick and error-free quality inspection process for the textile and apparel products.

As it is pointed out by most of the researchers, that replacement of manual quality inspection by the AI enabled quality inspection is very obvious. It is now established that AI enabled quality inspection systems are capable of effective quality control in shorter time-cycle and without much human interventions. Hence its free from human error and fatigue, Investment for AI enabled automated inspections systems are quite capable of ensuring satisfactory ROI.


  1. Nath, Sayoni & Dutta, Anirban. (2023). Application of Al in quality control in textile and apparel industry. Asian Textile Journal. Vol.32 , No.1. 52-59.
  2. Manisha Sahu, Analytic Steps, Sep 06, 2021, available at Al in The Textile Industry Applications and Impact Analytics Steps
  3. Grand View Research, 2017-2020, available at Global Textile Market Size & Share Report, 2022-2030 (grand view
  4. Available at
  5. Tesfay Welamo, Deng Sanpen, 2021. Available at 4 Application of Al IN TEXTILE JETIR 4th Oct 2022 2021 JETIR October 2021, Volume 8, Issue 10 (ISSN-2349-5162 -Search (
  6. SS Doke, N Sanmugam, Artificial intelligence and it’s application in Textile, Asian Journal, Jan 2002, Available at Artificial intelligence_and_its_applications_in_textiles
  7. Mr Tesfay Welamo, Journal of Emerging Technologies and Research, October 2021, Available at papers/JETIR2110307.pdf
  8. GM Nasira, P Banumathi, Asian Journal of Information Technolgy, January 2014. Available at https:// 286404580_An_ intelligent system__ for automatic fabric_inspection
  9. Application of Artificial Intelligence, Textile Blog August 2020, Available at
  10. Raghav Bharadwaj. November 2019, Artificial Intelligence in the Textile Industry-Current and Future Applica- tions, EMERI, Available at: https:// cial-intelligence-in-the-textile-indus-try-current-and-future-applications/

You may also like:

  1. Industry 4.0 for Textile and Apparel Industry
  2. Applications of Artificial Intelligence (AI) in Fashion Industry
  3. Augmented Reality (AR) in Fashion Industry
  4. Technological Factors Affecting the Fashion Industry
  5. Application of Internet of Things (IoT) in Textile and Fashion Industry
  6. Artificial Intelligence (AI) in Textile and Apparel Industry

Share this Article!

Leave a Comment