{"id":{"repo_id":"stellenbosch","oai_identifier":"oai:scholar.sun.ac.za:10019.1/136108"},"canonical_url":"https://search.dev.ndltd.org/etd/stellenbosch/oai:scholar.sun.ac.za:10019.1/136108","repository":{"repo_id":"stellenbosch","name":"Stellenbosch University","base_url":"https://scholar.sun.ac.za/server/oai/request"},"display":{"title":"A Computer Vision and Generative AI Framework for Trend Extraction and Brand-Aligned Fashion Design","abstract":"The fashion industry is renowned for its fast-paced and ever-evolving clothing lines, with new trends constantly emerging. Social media drives these rapid changes through the instant dissemination of trends and its global reach. The fashion industry is highly competitive, putting pressure on mass-market retailers to outperform their competitors. This thesis presents a novel framework to help retailers shorten the design-to-market turnaround time. The key concept is to use computer vision and unsupervised clustering techniques to identify trends from source images. It combines these trends with the aesthetics of a targeted brand and then utilises generative artificial intelligence to generate several trendy, brand-aligned design options. The framework has three components: an image classifier, a trend identification module, and a generative design module. The image classifier captures stylistic attributes from fashion runway photographs, the targeted trend source. These attributes include garment classes derived from a pre-trained YOLOv11s-seg model, colour palettes extracted using K-Medoids clustering, and fabric patterns classified using a pretrained ResNet34 model. The predictions are then fed into the trend identification module, which uses a self-organising map to identify trends. The trends, a secondary fabric dataset, and a sales catalogue of a targeted brand are processed in the generative design module. This module uses a novel mask-style blending framework to generate the designs. These results were validated through consumer surveys, interviews, and fashion trend prediction reports. The interviews revealed the value of this framework, and the researcher assessed the key advantages the framework presents over generic large language models.","abstract_html":"The fashion industry is renowned for its fast-paced and ever-evolving clothing lines, with new trends constantly emerging. Social media drives these rapid changes through the instant dissemination of trends and its global reach. The fashion industry is highly competitive, putting pressure on mass-market retailers to outperform their competitors. This thesis presents a novel framework to help retailers shorten the design-to-market turnaround time. The key concept is to use computer vision and unsupervised clustering techniques to identify trends from source images. It combines these trends with the aesthetics of a targeted brand and then utilises generative artificial intelligence to generate several trendy, brand-aligned design options. The framework has three components: an image classifier, a trend identification module, and a generative design module. The image classifier captures stylistic attributes from fashion runway photographs, the targeted trend source. These attributes include garment classes derived from a pre-trained YOLOv11s-seg model, colour palettes extracted using K-Medoids clustering, and fabric patterns classified using a pretrained ResNet34 model. The predictions are then fed into the trend identification module, which uses a self-organising map to identify trends. The trends, a secondary fabric dataset, and a sales catalogue of a targeted brand are processed in the generative design module. This module uses a novel mask-style blending framework to generate the designs. These results were validated through consumer surveys, interviews, and fashion trend prediction reports. The interviews revealed the value of this framework, and the researcher assessed the key advantages the framework presents over generic large language models.","abstract_has_math":false,"creators":["Nel, Chane"],"institution":"Stellenbosch : Stellenbosch University","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Burger, L. E.","Taljaard-Swart, Hanri"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026-03","date_published":"2026-03","updated_at":"2026-07-24T04:40:12Z","subjects":[],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://scholar.sun.ac.za/handle/10019.1/136108","outbound_label":"Repository record","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Burger, L. E.","Taljaard-Swart, Hanri"]},{"key":"dc:contributor.other","label":"Dc Contributor Other","values":["Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering."]},{"key":"dc:creator","label":"Author","values":["Nel, Chane"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-04-22T11:53:19Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-04-22T11:53:19Z"]},{"key":"dc:date.issued","label":"Date","values":["2026-03"]},{"key":"dc:publisher","label":"Institution","values":["Stellenbosch : Stellenbosch University"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://scholar.sun.ac.za/handle/10019.1/136108"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (MEng)--Stellenbosch University, 2026.","Nel, C. 2026. A Computer Vision and Generative AI Framework for Trend Extraction and Brand-Aligned Fashion Design. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/9f749abb-dc5b-4f2e-a170-f00e13dc084d"]},{"key":"dc:description.abstract","label":"Abstract","values":["The fashion industry is renowned for its fast-paced and ever-evolving clothing lines, with new trends constantly emerging. Social media drives these rapid changes through the instant dissemination of trends and its global reach. The fashion industry is highly competitive, putting pressure on mass-market retailers to outperform their competitors. This thesis presents a novel framework to help retailers shorten the design-to-market turnaround time. The key concept is to use computer vision and unsupervised clustering techniques to identify trends from source images. It combines these trends with the aesthetics of a targeted brand and then utilises generative artificial intelligence to generate several trendy, brand-aligned design options. The framework has three components: an image classifier, a trend identification module, and a generative design module. The image classifier captures stylistic attributes from fashion runway photographs, the targeted trend source. These attributes include garment classes derived from a pre-trained YOLOv11s-seg model, colour palettes extracted using K-Medoids clustering, and fabric patterns classified using a pretrained ResNet34 model. The predictions are then fed into the trend identification module, which uses a self-organising map to identify trends. The trends, a secondary fabric dataset, and a sales catalogue of a targeted brand are processed in the generative design module. This module uses a novel mask-style blending framework to generate the designs. These results were validated through consumer surveys, interviews, and fashion trend prediction reports. The interviews revealed the value of this framework, and the researcher assessed the key advantages the framework presents over generic large language models."]},{"key":"dc:title","label":"Title","values":["A Computer Vision and Generative AI Framework for Trend Extraction and Brand-Aligned Fashion Design"]}]}],"canonical_facts":{"dc:contributor.advisor":["Burger, L. E.","Taljaard-Swart, Hanri"],"dc:contributor.other":["Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering."],"dc:creator":["Nel, Chane"],"dc:date.accessioned":["2026-04-22T11:53:19Z"],"dc:date.available":["2026-04-22T11:53:19Z"],"dc:date.issued":["2026-03"],"dc:description":["Thesis (MEng)--Stellenbosch University, 2026.","Nel, C. 2026. A Computer Vision and Generative AI Framework for Trend Extraction and Brand-Aligned Fashion Design. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/9f749abb-dc5b-4f2e-a170-f00e13dc084d"],"dc:description.abstract":["The fashion industry is renowned for its fast-paced and ever-evolving clothing lines, with new trends constantly emerging. Social media drives these rapid changes through the instant dissemination of trends and its global reach. The fashion industry is highly competitive, putting pressure on mass-market retailers to outperform their competitors. This thesis presents a novel framework to help retailers shorten the design-to-market turnaround time. The key concept is to use computer vision and unsupervised clustering techniques to identify trends from source images. It combines these trends with the aesthetics of a targeted brand and then utilises generative artificial intelligence to generate several trendy, brand-aligned design options. The framework has three components: an image classifier, a trend identification module, and a generative design module. The image classifier captures stylistic attributes from fashion runway photographs, the targeted trend source. These attributes include garment classes derived from a pre-trained YOLOv11s-seg model, colour palettes extracted using K-Medoids clustering, and fabric patterns classified using a pretrained ResNet34 model. The predictions are then fed into the trend identification module, which uses a self-organising map to identify trends. The trends, a secondary fabric dataset, and a sales catalogue of a targeted brand are processed in the generative design module. This module uses a novel mask-style blending framework to generate the designs. These results were validated through consumer surveys, interviews, and fashion trend prediction reports. The interviews revealed the value of this framework, and the researcher assessed the key advantages the framework presents over generic large language models."],"dc:identifier.uri":["https://scholar.sun.ac.za/handle/10019.1/136108"],"dc:language.iso":["en"],"dc:publisher":["Stellenbosch : Stellenbosch University"],"dc:title":["A Computer Vision and Generative AI Framework for Trend Extraction and Brand-Aligned Fashion Design"],"dc:type":["Thesis"]},"updated_at":"2026-07-24T04:40:12Z"}