18599569. SYNTHETIC POSITIVE IMAGE GENERATION FOR FINE GRAIN IMAGE SIMILARITY BASED APPAREL SEARCH simplified abstract (Tata Consultancy Services Limited)
Contents
SYNTHETIC POSITIVE IMAGE GENERATION FOR FINE GRAIN IMAGE SIMILARITY BASED APPAREL SEARCH
Organization Name
Tata Consultancy Services Limited
Inventor(s)
Biswanath Pati of Bhubaneswar (IN)
Aravind Selvaraj of Chennai (IN)
Jayanta Mukherjee of Kolkata (IN)
SYNTHETIC POSITIVE IMAGE GENERATION FOR FINE GRAIN IMAGE SIMILARITY BASED APPAREL SEARCH - A simplified explanation of the abstract
This abstract first appeared for US patent application 18599569 titled 'SYNTHETIC POSITIVE IMAGE GENERATION FOR FINE GRAIN IMAGE SIMILARITY BASED APPAREL SEARCH
The abstract of the patent application describes a method and system for generating training data for a deep learning model to perform similarity search in an apparel search context.
- Positive and negative images are generated from each query image to create training data.
- The training data is used to train a deep learning model for similarity search.
- This approach eliminates the need for a large amount of training data per Stock Keeping Unit (SKU).
- Domain experts may not be required for apparel classification to generate training data.
- The deep learning model can efficiently perform similarity search in a computationally lighter manner.
Potential Applications: - E-commerce platforms for apparel search - Visual search engines for fashion industry - Recommendation systems for clothing items
Problems Solved: - Shortage of training data for deep learning models in apparel search - Heavy computational load in performing similarity search - Dependency on domain experts for apparel classification
Benefits: - Improved accuracy in similarity search - Reduced computational resources required - Faster and more efficient apparel search process
Commercial Applications: Title: "Enhanced Apparel Search Technology for E-commerce Platforms" This technology can be used by e-commerce platforms to enhance their apparel search capabilities, leading to better user experience and increased sales. The market implications include improved customer satisfaction, higher conversion rates, and competitive advantage in the online retail industry.
Questions about Enhanced Apparel Search Technology: 1. How does this technology improve the efficiency of apparel search in e-commerce platforms? - This technology generates training data efficiently for deep learning models to perform similarity search, reducing the computational load and improving accuracy. 2. What are the potential applications of this technology beyond e-commerce platforms? - This technology can also be used in visual search engines for the fashion industry and recommendation systems for clothing items.
Original Abstract Submitted
In apparel search context, process of finding a similar item out of thousands of other items is a cumbersome and computationally heavy process. In order to build a deep learning model that can perform the similarity search, hundreds of training images per Stock Keeping Unit (SKU) are required. Due to shortage of training data, this approach fails to generate a deep learning model that can perform the similarity search in intended manner. The existing approaches may also require domain experts to perform classification of apparels, so as to generate the training data. The method and system disclosed herein provide an approach in which positive images and negative images are generated from each query image, which in turn are used for generating a training data. The training data is then used to generate a deep learning model, which is used to perform the similarity search.
