I work at ivi.ru helping to build computer vision algorithms and services. I'm interested in deep neural networks and try to apply them to video summarization, its temporary segmentation, and tagging. Also, I want to try an audio analysis. For example speech recognition or music generation. Luckily, this is connected to my other activity.
You can usually find me on the internet as @sashulyak .k H @
How We Learned How to Divide Videos Into Scenes Using Tricky Math (RU) — At ivi.ru I have implemented the algorithm, which divides videos into semantic scenes. This article is about how I accomplished it and made an evaluation of the speed and accuracy of the algorithm. The source code of the implementation is available on Github .
How I got inspired by Half-Life and Dan Simmons — and recorded an album (RU) — The story about how my first music album was made. I describe all the stages from idea to mastering and release.