7 November 2016 in International Journal of Clothing Science and Technology
Purpose: With the developments of e-commerce markets, novel recommendation technologies are becoming an essential part of many online retailers’ economic models to help drive online sales. Initially, the purpose of this paper is to undertake an investigation of apparel recommendations in the commercial market in order to verify the research value and significance. Then, this paper reviews apparel recommendation techniques and systems through academic research, aiming to acquaint apparel recommendation context, summarize the pros and cons of various research methods, identify research gaps and eventually propose new research solutions to benefit apparel retailing market. Design/methodology/approach: This study utilizes empirical research drawing on 130 academic publications indexed from online databases. The authors introduce a three-layer descriptor for searching articles, and analyse retrieval results via distribution graphics of years, publications and keywords. Findings: This study classified high-tech integrated apparel systems into 3D CAD systems, personalised design systems and recommendation systems. The authors’ research interest is focussed on recommendation system. Four types of models were found, namely clothes searching/retrieval, wardrobe recommendation, fashion coordination and intelligent recommendation systems. The forth type, smart systems, has raised more awareness in apparel research as it is equipped with advanced functions and application scenarios to satisfy customers. Despite various computational algorithms tested in system modelling, existing research is lacking in terms of apparel and users profiles research. Thus, from the review, the authors have identified and proposed a more complete set of key features for describing both apparel and users profiles in a recommendation system. Originality/value: Based on previous studies, this is the first review paper on this topic in this subject field. The summarised work and the proposed new research will inspire future researchers with various knowledge backgrounds, especially, from a design perspective.
3 June 2019 in International Journal of Clothing Science and Technology
Purpose: The big challenge in apparel recommendation system research is not the exploration of machine learning technologies in fashion, but to really understand clothes, fashion and people, and know what to learn. The purpose of this paper is to explore an advanced apparel style learning and recommendation system that can recognise deep design-associated features of clothes and learn the connotative meanings conveyed by these features relating to style and the body so that it can make recommendations as a skilled human expert. Design/methodology/approach: This study first proposes a type of new clothes style training data. Second, it designs three intelligent apparel-learning models based on newly proposed training data including ATTRIBUTE, MEANING and the raw image data, and compares the models’ performances in order to identify the best learning model. For deep learning, two models are introduced to train the prediction model, one is a convolutional neural network joint with the baseline classifier support vector machine and the other is with a newly proposed classifier later kernel fusion. Findings: The results show that the most accurate model (with average prediction rate of 88.1 per cent) is the third model that is designed with two steps, one is to predict apparel ATTRIBUTEs through the apparel images, and the other is to further predict apparel MEANINGs based on predicted ATTRIBUTEs. The results indicate that adding the proposed ATTRIBUTE data that captures the deep features of clothes design does improve the model performances (e.g. from 73.5 per cent, Model B to 86 per cent, Model C), and the new concept of apparel recommendation based on style meanings is technically applicable. Originality/value: The apparel data and the design of three training models are originally introduced in this study. The proposed methodology can evaluate the pros and cons of different clothes feature extraction approaches through either images or design attributes and balance different machine learning technologies between the latest CNN and traditional SVM.
28 January 2014 in IEEE Transactions on Automation Science and Engineering
Three-dimensional (3D) body scanning technology opens opportunities for virtual try-on and automatic made-to-measure apparel design. This paper proposes a new feature-based parametric method for modeling human body shape from scanned point clouds of a 3D body scanner [TC]2. The human body model consists of two layers: the skeleton and the cross sections of each body part. First, a simple skeleton model from the body scanner [TC]2 system has been improved by adding and adjusting the position of joints in order to better address some fit issues related to body shape changes such as spinal bending. Second, an automatic approach to extracting semantic features for cross sections has been developed based on the body hierarchy. For each cross section, it is described by a set of key points which can be fit with a closed cardinal spline. According to the point distribution in point clouds, an extraction method of key points on cross sections has been studied and developed. Third, this paper presents an interpolation approach to fitting the key points on a cross section to a cardinal spline, in which different tension parameters are tested and optimized to represent simple deformations of body shape. Finally, a connection approach of body parts is proposed by sharing a boundary curve. The proposed method has been tested with the developed virtual human model (VHM) system which is robust and easier to use. The model can also be imported in a CAD environment for other applications.