Streaming is one of best forms of entertainment for gamers, casual and hardcore alike – and one of Spencer’s favorite games to watch is Escape From Tarkov. However, while streams can be fun to watch, they often last several hours, and not everyone has the time to sit through hours of gameplay to catch the most thrilling moments.

In this session, we'll dive into how Spencer used machine learning, code, and media processing tools to identify individual games within 8+ hour streams – from knowing the map being played on to the kill count within the game to knowing exactly when the most intense moments occurred. We'll explore how this data is analyzed, timestamped, stored, and ultimately used to cut the raw stream into individual, watchable segments of gameplay.

This engaging session will spotlight a series of technologies that was used to ultimately solve the problem: how can I watch the best games, on my favorite maps, for my favorite streamer, LVNDMARK? If you're a developer interested in machine learning, AI, data processing, or media analysis, this session will leave you with insights, inspiration, and a new perspective on the powerful tools we developers have available to us.