An artificial intelligence system developed at IIT Roorkee has demonstrated the ability to translate detailed written clothing descriptions into professional fashion sketches by linking design language directly with visual output. The framework enables AI models to interpret specialised garment terms related to cut, material, texture and style. The capability supports faster and more precise design workflows in fashion and related creative industries.
- The Fashion Language Outfit Representation for Apparel Generation (FLORA) dataset pairs professional fashion sketches with textual descriptions, allowing AI models to learn connections between design language and corresponding visual features.
- The Nonlinear Expressive Representation Adapter (NeRA) improves how AI models adapt to complex and nuanced fashion vocabulary, enabling faster learning and more accurate generation of high-fidelity sketches from text inputs.
- The study, titled ‘Dressing the Imagination: A Dataset for AI-Powered Translation of Text into Fashion Outfits and a Novel NeRA Adapter for Enhanced Feature Adaptation’, was led by Prof Sparsh Mittal at IIT Roorkee and announced by the Press Information Bureau recently.
THE STUDY: The research centres on the development of a fashion-specific artificial intelligence framework that links detailed textual descriptions with professional design sketches rather than product photographs. It introduces a paired dataset and a specialised adapter to help AI models interpret complex garment vocabulary used by designers. The work represents India’s first structured effort to translate professional fashion language directly into sketch-based visual outputs.
- The dataset pairs professional fashion sketches with detailed textual descriptions using terminology related to silhouette, material, texture and overall style.
- Unlike existing fashion datasets based largely on catalogue or shopping images, the dataset reflects how designers actually conceptualise and sketch garments during the design process.
DESIGN IMPLICATIONS: By enabling AI to understand professional fashion vocabulary, the framework reduces the time and effort required to convert design ideas into usable visual sketches. This supports faster iteration during early design stages and lowers dependence on manual sketching for routine concept development.
- Designers can describe garments using detailed technical language and receive sketch outputs that reflect intended silhouettes, materials and styling choices.
- The system shortens design cycles by automating parts of concept visualisation without replacing the creative decision-making process.
- Improved accuracy in early-stage sketches can reduce downstream revisions, supporting more efficient development processes.
WIDER CONTEXT: The development positions IIT Roorkee’s work within a growing push to apply artificial intelligence to creative and design-led industries in India. By focusing on indigenous datasets and design processes, the research aligns technical innovation with local industry needs rather than imported design conventions. The approach supports wider efforts to integrate AI into sustainable manufacturing, digital design and creative technology ecosystems, while aligning with the United Nations Sustainable Development Goals, including SDG 9 on industry, innovation and infrastructure and SDG 12 on responsible consumption and production.
- The open-source release of the dataset and adapter allows designers and researchers to build on the work without reliance on proprietary tools.
- The framework aligns with national initiatives such as Digital India, Make in India and the National AI Mission by supporting locally developed technology.
- The research also connects AI innovation with sustainability goals by enabling more efficient and precise early-stage design practices.
WHAT THEY SAID
Our work brings AI and design together to make creativity easier and faster. Modern fashion includes small details that many AI models could not understand well. With FLORA and NeRA, we help AI understand designers’ ideas through detailed sketches and precise design terms. This can support developments in fashion, retail and other areas.
— Prof Sparsh Mittal
Department of Electronics and Communication Engineering
Indian Institute of Technology Roorkee
This remarkable achievement underscores IIT Roorkee’s commitment to cutting-edge interdisciplinary research. Innovations like FLORA and NeRA demonstrate how AI can be harnessed to enhance India’s creative economy and bolster its global leadership in design and technology.
— Prof Kamal Kishore Pant
Director
Indian Institute of Technology Roorkee
