Back in 2006, while working as a graduate researcher at University of Pittsburgh (USA), Amruta first realized the potential of applying AI / ML algorithms to process the audio-visual content in television programs and movies. Her paper on Humor Analysis in FRIENDS tv show was published in one of the top international conferences in Natural Language Processing area at EMNLP-2006 in Sydney.
Her interests to work in this field were further reinforced when she was offered an opportunity to work with SONY at their headquarters in Tokyo, Japan. It was during this internship that Amruta realized the potential industry applications of using AI / ML algorithms in media and entertainment domain.
What fascinated Amruta the most was the variety of entertainment programs and online videos out there which can be utilized for research and experimentation – news broadcasts, sports, drama, action, comedy, fashion shows, music videos, talk-shows, advertisements, tv commercials, documentaries on science, history, travel, wild-life and so much more..
The opportunity was enormous – what we could possibly create from all this data! Even more so, when we consider the online distribution & streaming platforms like YouTube, NetFlix, Amazon Prime, most of which offer SDKs & APIs for developers to create their own applications and plug-ins. For Amruta, this was a playground for testing her creative ideas and an opportunity to integrate stand-alone applications with existing service providers..
Recent years have experienced tremendous progress in the areas of Deep Learning, Machine Learning, Computer Vision & AI, from academic research to more practical industry oriented applications. Free access to open-source libraries and robust deep-learning platforms (like TensorFlow, PyTorch, Keras) have made it possible even more so than before to implement these algorithms more efficiently and accurately on GPUs & Cloud.
Over the past few years, Amruta has created several prototypes, demos & proposals for various clients to showcase potential applications of AI / ML in media and entertainment domain, including – fashion, advertising, sports, music, action, comedy, visual effects, video games & animation.. While the company was officially incorporated in May 2019, some of the projects and ideas listed here date back to as early as 2006, and demonstrate over a decade of thought process and experimentation efforts..