Turing Analytics: Transforming Retail With AI-Driven Visual Search and Recommendations

Turing Analytics is an innovation-focused startup in the machine learning and computer vision domain that helps retailers enter into the world of Visual Commerce through its Visual Intelligence solutions. With powerful machine learning capabilities, the startup trains its algorithms to understand images like humans, identify what objects they contain, tag them accordingly and use visual search to find similar images and objects. The company is teaching machines to see the world the way humans do. The solutions help retailers improve product discovery, customer engagement and boost conversion rates.

Turing Analytics offers the following products:

•  Visual Search: Enable shoppers to search items using photos of products they are inspired to buy.

•  Visual Recommendations: Suggest products on E-commerce platforms based on visual similarity.

•  Identification and Tagging: Automatically detect objects and generate relevant tags for all the images.

•  Visual Cart: Scan and shop solution for retailers. Easy, cart-less and hassle-free next-generation shopping experience for customers.

These products help retailers to use customer images and videos from both offline and online sources to understand customer opportunities at a deep granular level, real-time analytics, predict trends, improve customer engagement by enabling Omni-channel retail, amongst others.


Enabling Retailers Provide Superior Customer Experience

With its computer vision technology powered suite of products, Turing Analytics aims to deliver smart solutions that enable top retailers of the world to provide a superior shopping experience to their customers.

The founders of the company were tech enthusiasts looking for an opportunity to start an innovation-driven company. In 2015, they were convinced that Machine Learning (ML) is the technology of the future, and decided to create products that would solve business problems using ML.

In August 2015, they identified that Indian e-commerce websites were not able to unlock the full potential of customer reviews as off-the-shelf NLP engines were not good enough in the Indian context. Hence, the leaders created an NLP engine that worked for Indian English as well as Hinglish customers. This innovation was used to create a “Customer Review Analytics System”.  Shopclues was the first customer on-boarded in January 2016 and the company has been working with them since then.

From there on, the founders moved on to building image/video recognition solutions. The first solution was developed for Tata GTIO (for Tata Trent) which was a fashion trend recognition system. The system predicted the sales ability of a product with just image of the product as input with 85% and above accuracy.

The company has since moved on to build a complete visual intelligence suite.


Meet the Passionate Leaders

The company is co-founded by Aditya Patatdia and Divyesh Patel. Divyesh is an ambitious person who has a very clear vision of the life he wants to lead. He makes realistic long-term plans and acts on them pragmatically, starting a tech company was always part of the plan.

At the time, Divyesh was pursuing his MBA in Marketing from SCMHRD he used to take a keen interest in statistics and market research methodology. That coupled with his background in Computer Science enhanced his ability to grasp the latest research in Machine Learning and implement it in code. While Divyesh was working from his dorm room, Aditya who is also an engineer and holds an MBA from FMS, New Delhi was working extra hours post his job to build the first product. After months of hard work, the pair launched their first product in December 2015.

Divyesh loves traveling, swimming and discussing philosophy. He believes that building a long-term sustainable business is a crucial step towards a larger goal of living a free and fulfilling life that would leave a mark on time.


Pioneering Indian E-commerce Success

 Turing Analytics is among a few pioneers of deep learning tech in India that has been working on adapting this tech for Indian use cases since 2015. For example, the company started by making the first ever “Customer Review Analytics” system for the Indian e-commerce industry which could understand the reviews written in Hinglish. This was a real breakthrough as about 20-30% of reviews coming in at the dawn of E-commerce in India were in Hinglish. The system offered a way to moderate the reviews reliably and automatically.

Since then the focus has been to make sure that the best AI tech available in the world works for Indian use cases. It is part of the Turing Analytics core mission. In future, the company wants to go beyond just implementation of deep learning and start doing a fundamental search in the field. There is a lot of data and a lot of challenges in India to enable this.


Leading the Competition with Disruptive Innovation

Divyesh feels the biggest impact of innovations in the industry is that they have made machine intelligence commercially viable. Things that could be done only in the R&D departments of big corporates can now be done in handheld devices by anyone. Machine intelligence has become a commodity, and it is already affecting our day-to-day lives.

Turing Analytics implements state-of-the-art deep learning technology from scratch and fundamental research about new neural network architectures published by big companies like Google or Microsoft. However, those research papers only address the well-established academic problems like ICLR (The ImageNet challenge).

“To understand that research and implement it for practical problems on real-world data is a different ball game altogether. For example, let’s say you want to train a neural network model that understands the difference between western dresses and Indian Kurtis. One has to start from scratch – defining what characteristics are important to make this distinction, manually collect the data from various places, run about 50 different experiments to achieve the right mix of architecture plus training parameters which lead to the best accuracies,” Divyesh said.

And after doing all of these, you will realize that there is a whole line of clothing called fusion wear for which your model will never work! And it’s this process, the learning curve, where domain experts and engineers work together to achieve perfection which makes Turing Analytics innovative.

Turing Analytics’ clients like Tata GTIO and Future Group C&D labs are the most important partners in this process. It is owing to their trust and belief for cutting-edge tech the company was able to carry out so many experiments.


Awards and Testimonials

 The company has been listed as “Emerging Analytics Startup- 2018 by Analytics India Magazine. Below are some of the encouraging feedback received from the customers:

 “We want to thank you for the successful implementation of the Fashion Trend Scoring system using deep learning that you have developed for us. This software product is capable of determining the trendiness score of a shirt or t-shirt using advanced neural network algorithms… We hope to work with your team again on upcoming projects. Thanks again for the job well done.”
GTIO Tata Sons Ltd.

 “We, at C&D Labs, the Tech and data Innovation arm of Future Group, on-boarded Turing Analytics in 2017 as part of our Ecosystem outreach program. They are one of the 50 other startups we have considered for retail disruptive tech co-creation. Besides FBB online, we are considering Turing for various other image analytics use cases for other formats of Future Group” said Somu Vadali, Head of C&D labs.


Overruling the Challenges

The biggest challenge the company faced so far was to build a visual search engine which can search across more than 5 million images within a second. It required some Google level engineering feats, which involved tinkering at the most fundamental levels of computation.

Besides the engine, Turing Analytics also had to train a massive neural network which could understand the nuances and differences about products from very diverse categories like clothing, furniture and FMCG. It takes a lot of time to collect, clean and standardize data for such tasks. The company has about a million images in a proprietary database which is used in training AI models.
Also, for a startup with a tech team smaller than 10 people, it is a big achievement.


Foresight for Years Ahead

Divyesh believes that it is just the beginning of a revolution. AI has the power to make every aspect of our lives more agile & more productive. It is going to transform the process in businesses and governments. AI will remove a lot of bottlenecks, and unleash faster collaborations. As more things get done in less time, the progress of the human society will only accelerate.

According to Gartner, by 2021, early adopter brands that redesign their websites to support visual and voice search will increase digital commerce revenue by 30%.

As for Turing Analytics, the future is about expansion across geographies and then business verticals. Turing Analytics is in process of raising its first institutional funding round. The company aims to expand its business 10 times in the next couple of years.

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.

Your data will be safe!Your e-mail address will not be published. Also other data will not be shared with third person.