IIT-R Unveils India’s First Text to Fashion Sketch AI System
Roorkee, India – Researchers at the Indian Institute of Technology Roorkee (IIT-R) have made a groundbreaking advancement in artificial intelligence by developing a system that generates high-fidelity fashion sketches directly from textual descriptions. This innovative technology marks a significant milestone as it is the first of its kind in India.
The Need for a New Approach
Traditionally, most AI applications in the fashion industry have relied heavily on photographs and catalog images. While these applications have proven effective for online retail, product searches, and recommendations, they often fall short in capturing the creative and conceptual processes that fashion designers employ. Designers typically conceptualize their ideas through sketches and use industry-specific terminology, making it challenging for existing AI tools to interpret complex design descriptions beyond basic keywords.
Introducing FLORA
To address these limitations, the researchers at IIT-R developed FLORA (Fashion Language Outfit Representation for Apparel Generation). This is the country’s first comprehensive fashion sketch dataset specifically designed for text-to-outfit generation. The dataset consists of 4,330 curated pairs of professional fashion sketches and their corresponding textual descriptions. These descriptions utilize terminology commonly used by designers, capturing intricate details such as silhouette, fabric, texture, construction, and overall style.
Why FLORA is Unique
Unlike conventional photo-based datasets, FLORA reflects the actual sketches that designers create in their sketchbooks. This allows AI models trained on FLORA to interpret not only simple prompts like “a blue dress” but also more complex descriptions such as “a cobalt blue evening gown with a softly draped cowl neckline and a ruched bodice.” This capability enables more precise and expressive design generation at the conceptual stage.
Overcoming Limitations in AI Training
However, the researchers noted that having a dataset alone is not sufficient to achieve accurate results. Existing parameter-efficient fine-tuning techniques, such as LoRA, LoKR, DoRA, and LoHA, often struggle to capture complex semantic relationships in design language. To overcome this limitation, the team introduced NeRA (Nonlinear low-rank Expressive Representation Adapter), a novel adapter architecture based on Kolmogorov-Arnold Networks.
How NeRA Works
Professor Sparsh Mittal, the principal investigator of the study from IIT Roorkee’s Mehta Family School of Data Science and Artificial Intelligence, explained, “NeRA uses learnable nonlinear transformations that help AI models better capture subtle stylistic intent, achieve stronger semantic alignment, and converge faster during training.” This innovative approach enhances the AI’s ability to interpret and generate fashion designs accurately.
Collaborative Research Effort
The research was a collaborative effort involving several experts, including:
- Gayatri Deshmukh (Independent Researcher)
- Somsubhra De (IIT Madras)
- Chirag Sehgal (Delhi Technological University)
- Jishu Sen Gupta (IIT BHU)
Potential Impact on the Fashion Industry
The framework developed by the IIT-R researchers has the potential to revolutionize design workflows in the fashion industry. By enabling greater personalization and supporting sustainable fashion practices, this technology can significantly reduce trial and error during the early stages of design. This is particularly important in an industry increasingly focused on sustainability and efficiency.
Conclusion
The unveiling of India’s first text to fashion sketch AI system represents a significant advancement in the intersection of technology and fashion design. As AI continues to evolve, tools like FLORA and NeRA will likely play a crucial role in shaping the future of fashion, providing designers with innovative ways to conceptualize and create their work.
Note: This article is based on the latest developments in AI technology as reported by the Times of India and aims to provide an overview of the advancements made by IIT-R in the fashion industry.
