There are pretty much advanced and interesting technologies in the world, which are now helping people to solve business and other requirements, including research. There are couple of these technologies, to list down some, AI, Computer Vision, Speech Recognition etc.
We are identified as a challenge taker for these technologies, breaking technical barriers and finding solutions for customers. Below are some of the things we have done.
We won 2nd place at MIT Global Startup Labs (SL) a competition organized by Massachusetts Institute of Technology – USA. We presented project EagleEye, one of fastest real-time human movement recognition surveillance systems, possibly the fastest by the year of 2014. We took live feeds from 5 surveillance cameras, recognized human only movements, took actions based on the movements we detect. We did all of these with just 2GB of RAM. The target of this project was to solve the issue of false alarms caused by existing surveillance systems like motion sensors and heat sensors, because the “humans” are the real intruders a surveillance system should respond to, not to falling leafs, passing by pets etc.
The project went viral and reached several organizations, countries and even into tech giants. The project was subjected to 3 patent-able technologies.
Before Google Cloud introduce the service of real time speech recognition and when IBM technology was less accurate, we had to develop an API for real time speech recognition, for Android. Using the existing
speech to text technology in android, we managed to overcome the technical barrier of making it real time. It was amazing amount of research, test and fails which finally brought everything to success.
We made a small demo for our client to show the working app, check below.
We were given challenge-full tasks from our clients all the time. We have been asked to develop a system to search for a specific string in a MySQL database of over 10 million records containing the web scraped information from all around the world, in 5 seconds with only a 512 MB of RAM in server. With the support of the customer several methods has been tried including High Performance computing.
Finally we came across a solution which solved the issue, searching over 10 million string records in 1/10 of a second. The system was scallable up to couple of billions records.