A Delhi based Artificial Intelligence startup, Staqu is nothing short of a revolutionary idea that is taking fashion e-retail and e-commerce industry to a new level. Founded in 2015, the company has pioneered a bidirectional image understanding technology, an innovative image-to-image matching system, that simplifies the image search, automated meta-tag generation and real time product recommendations.
To know more about the company and understand the underlying technology, Analytics India Magazine interacted with Atul Rai, CEO & Co-founder at Staqu, who has some interesting facts to reveal. Let’s read here-
On re-defining search using Artificial Intelligence-
Staqu is all about artificial intelligence, and the most interesting part is that it is using AI to revolutionize search engines. “Whenever we think about search engines, we have a pre-defined picture in our mind, of a textual box where one can type a text query and get results in some textual form”, says Rai, who has been quite instrumental in Staqu’s development such as winning the IBM’s Global Entrepreneur Program and raising the initial round of investment from Indian Angel Network.
“But if you see the current trend of data on internet, you will find that most of it is now flowing in forms of images – 70 per cent of data on the internet is in the form of images. This, in turn, means that the quest for image search related solutions will increase in the near future”, he adds.
Though images are the next go-to thing, the underlying problem with images is that one cannot define every image in terms of text. Rai explains “Let’s say, if you want to search for “Alia Bhatt’s lehenga from Badrinath ki Dulhaniya” it’s really difficult to convert every colour and pattern in form of text and perform search in some traditional search box. This is one example, where searching through images simplifies things.”
Tapping onto this idea of utilizing domain specific meaningful information from images and using it for different purposes, not limited to search, Staqu was founded in 2015 and has since been “re-defining search using AI”.
On the idea behind using AI-
In computer vision, image-to-image search has been a much known problem, and researchers have been actively working in this domain since the 80’s. And it was with the rise of deep learning based approach post 2013, that it has been possible to extract powerful visual features from images that has led to some great results in the said field.
“Previously, I was working with one such similar problem (image to image search) with one of the UK based universities. Once I completed my research, I decided to expand this idea further to solve some known problems in e-commerce and that’s where I met and collaborated with my fellow co-founders, Anurag, Chetan and Pankaj, who have diverse knowledge and skills needed to democratise this technology”, adds Rai, who has a Master’s in Artificial Intelligence from the University of Manchester. Rai also has won the prestigious 21st European Image processing challenge: SSIP 2013 and has the Best Research Paper Aware in DICTA 2014, IEEE Australia, to his credit.
On “the Fashin App”-
Adding a note on the products and services offered by Staqu, Rai says “Staqu is currently working in two domains – B2B and B2C. In the B2B domain, we provide two suites of products, one is for mobile companies (OEM), where we wrap our AI technology to develop Ad tech platforms and other is for E-commerce companies, where we provide API based solutions.”
The company has developed a first consumer centric mobile application “Fashin” (Available for both Android and iOS) for fashion e-commerce domain that aims to provide one stop solution to discover and search fashion products across the e-commerce websites in the least possible time.
“The mobile application assists users to not only search the product through its image search feature but also helps the users to get wardrobes of their favourite Bollywood celebrities directly from YouTube videos”, says Rai.
Fashin app comes with features like image search, tag based filtering, real time trend analysis from video and a lot more. Once the app is downloaded, the user can stream latest Bollywood Videos and the application would extract the similar jacket or T-shirt available at various websites at different prices.
The app also allows the user to either click at that moment or upload an image from phone’s gallery and search for preferred fashion with best prices available online. It also enlists the best deals from the top-notch brands. It also allows to further simplify fashion discovery by taking a picture of favourite outfit and uploading it to browse for similar options available online.
On offering customized solutions-
Rai says “Different clients have different requirements. In the current competitive environment, each of our clients want an edge on other players hence customisation is one of the key components of Staqu’s product suite.”
“In the customisation process, we develop and wrap up our AI technologies in some form of API as well as product required for our clients. We develop plug and play AI powered APIs and SDKs which can be integrated in their existing products with minimal technical interference”, he adds.
Staqu currently works with brands like Paytm, Karbonn, Intex, Panasonic, Lava , Roposo , VistaPrint etc. for different set of AI products.
On the growth story-
Rai is quick to admit that within 11 months of their market launch, Staqu is not only working with some of the big brands in the industry but have also closed the contract for more than two years with some of them. “I believe our clientele tells better story than the number or figures I can provide in support of our technology”, he says.
“We started Staqu with the sole motto of democratising AI to the common audience and we already have launched “Fashin” which completely automates mobile application target for fashion loving audience. The app is consistently gaining traction and has more than 10K downloads without any marketing efforts and 800-900 daily active users with session of 3-4 minutes”, he says.
On a concluding note-
He concludes leaves a advice to the startup community by adding “Newcomers should focus on building a great product but should not miss out on making money out of it. As ultimately, building great product and scaling it would require money, so rather than running behind investors, you should try to generate it from the product itself and use it to innovate further. I would further urge aspiring entrepreneurs to never stop building and never stop innovating.”
Try deep learning using MATLAB