Model generated Watch designs
Generating watch images using Generative Adversarial Networks
One of India’s largest watchmakers wanted to create a Deep Learning model for creating new designs for watches. We created a Generative Adversarial Network (GAN) and added functionality to control the type of images generated by features selected by the user. The user can select various watch features like color of the band, color of the dial, gender and band material. Our advanced GAN combined a regular GAN with a multilabel-multiclass classification model to create image of a watch that matches the user’s selected criteria. This model combines human creativity with artificial intelligence to bring out the best of both.
Computer Vision for Measuring Compliance of Product Placement on Shelves (Planogram)
Automating process of measuring compliance of product placement on shelves (Planogram) at retail establishments using Computer Vision
A large consumer goods manufacturer wanted to measure compliance of their product placement on retail shelves by various retailers. They required to identify various products, count their stock and make sure they are arranged in right order. This information is used to calculate incentives that they pay to retailers for special shelf space. We created an object detection model that identifies various products from shelvess and used various algorithms to calculate compliance percentage with a model planogram. This saves significant amount of manual labor for manually verifying each stores planogram for the manufacturer.
Predictive Analytics for Driving Multi-million Dollar Marketing Strategy
Created a comprehensive marketing strategy based on multiple predicitve models for a US based travel company with multi-million dollar marketing budget
One of America’s top travel company for experiential travel, used direct marketing for its loyal customer base. Originally the marketing strategy was product focused, trying to identify right customers for each product. We created several predictive models to change the marketing strategy to be customer focused, identifying best suited products for each customer. This improved the effectiveness of each marketing campaign significantly, saving the company in marketing costs.
The models were created to identify customer’s lifetime value to the company, likelihood of purchasing various products and measuring their web browsing behaviour to identify where they are in the buying cycle. We combined online and offline data from sources like website, call center, past trips and surveys to build sophisticated models. Company experienced significant growth in travellers while the marketing costs reduced.