Let The Images Describe Themselves

CIOReview Team | Tuesday, 04 August 2015, 12:16 IST

AI based start up changing the way you perceive the world.

A picture is worth a thousand words. Staqu, a Delhi based AI startup, is crunching data and smoking GPUs to make it a reality for online consumers at multiple levels, such as search, trend-analysis and recommendation. Staqu explains the rationale behind using images for search by citing examples of complex patterns and colors appearing in clothes, tiles, curtains, furnitures etc. that are impossible to describe using a text-based query. Staqu's visual search module takes an image of a clothing from the users and search through a database of millions of images to return most visually similar results within seconds. Staqu plans to provides a real-time analysis of the fashion trends by crawling images of celebrities and stars from the internet to help the users become aware of the latest trends in fashion. Eventually, Staqu wants to develop an India-specific fashion engine,, that will take the image of a user and suggest combinations of fashion apparels to improve user looks. A similar idea will be used to revamp the current recommendation engines on e-commerce websites.

Abhishek, Co-founder Staqu, explains the vision of Staqu - “At Staqu, we want to contribute to the technology instead of just consuming it. Take for example, the recommendation systems used by the e-commerce websites for fashion amenities. Mostly these engines are based on the choices of previous users on the website. This approach is agnostic to the content of the item, latest trends, upcoming events, celebrities and most importantly the users themselves ! We approach recommendation systems from the user’s point of view and combine item’s content, current trends and user’s image, if user agrees to provide, to recommend matching fashion items to the user.”    

Atul, CEO and Co-founder, Staqu, says - “The internet has the biggest chunk of data dedicated to images and videos that need humans for annotation, content description and recommendation; at Staqu we are building intelligent engines that can extract the knowledge and insight from the multimedia content while minimizing the human effort involved. Let the images describe themselves!. Staqu is not only limited to e-commerce section, we aspire to provide our services to all the sectors that can benefit from computer vision, machine/deep learning and data analysis, which is virtually all the sectors. Staqu has a research team consisting of experts in biometrics, image/media analysis, human-computer interaction, safety analysis, energy systems etc. that allows  to expand rapidly in multiple sectors simultaneously.”

Staqu Research brings PhDs and faculties from MIT, University of Manchester, University of Texas, University of Maryland, Washington State University and IITs to develop a one-stop state-of-the-art lab for various automation tasks ranging from fashion industry to industrial automation. Prof. David W. Jacobs at University of Maryland, referred to as Prof. Dumbledore, is Staqu’s Research Director. David has led the research on the matching module for the pioneering and the most successful field-guide Leafsnap that has more than 1.5M users and voted among top 10 useful iPhone Apps  (video). David says - “Mobile phones and internet connections provides us a way to put the power of computer vision and machine learning in the hands of people to make their lives more comfortable and a better learning experience (video).”  

An awesome research requires a super-awesome design and technical team that can deploy the intimidating and mysterious algorithms in the form of a simple, intuitive and user-friendly application. Staqu has a strong and experienced technical and design team to interface the research with the consumer in a seamless manner. The team has expertise ranging from research experience to mobile platforms to server side software development. The tech team takes pride in counting some of the smart apps such as Blink based selfie, Gesture based selfie, achieving DSLR effect in an image without any additional hardware as its previous work-experience. The members of the team have experience of developing smart social and news aggregator, analytics libraries, app store, multitasking app etc. for big Indian mobile brands that are being used by millions of Indian mobile users.

Don't Miss ( 1-5 of 25 )