On February 21, 2005, one of the best animated series to ever grace the TV screen aired for the first time. Avatar: The Last Air Bender will always be one of my favorite animated TV shows from childhood. The show centers Aang, the avatar who can bend all four elements, in his quest to stop the imperialist Fire Nation in their conquest of the world. The plot seems simple enough but all is not as it seems. It is quite rare for a kid’s show to tackle themes geared towards more mature audiences like genocide, marginalization, oppression, female empowerment, stoicism and many more. No wonder ATLA has developed a cult following in recent years. In this blog post, I have utilised yet another pose estimation model called posenet to bend the four elements using body movements.

Two events inspired this post. On April 24, 2020, a new version of the Yolo object detection algorithm was released and on May 28, 2020, the president of the Philippines made it official that the capital, Metro Manila, would be placed under General Community Quarantine(GCQ) from the previous Modified Enhanced Community Quarantine(MECQ) amidst the Convid-19 Pandemic. In this blog post, I will demonstrate how Deep Learning can help in monitoring distancing behaviour of the general public.

Pose estimation is a computer vision technique that detects human figures in images through the localization of keypoints. These keypoints are usually the human joints and are connected to form a skeletal/stick representation of the subject. It has a variety of applications which include augmented reality, motion capture and robotics. In this blog post we will apply pose estimation to Kpop dances.

Managing machine learning projects can be a daunting task without the proper tools to assist you. As with traditional software engineering, versioning is an essential part of the machine learning workflow. However, unlike traditional software engineering, machine learning involves massive files, consisting of datasets and model artifacts, that current version control systems do not support. In addition, machine learning experiments are evaluated by a number of metrics that need to be tracked. In this blog post, I will walk you through my personal machine learning workflow.

When I was a kid, I used to watch an anime called Naruto. It was a show which followed the adventures of Naruto, the main protagonist who was a ninja. In naruto, ninjas possesed “chakra” that allowed them to perform ninja jutsus or techniques. To use these techniques, they had to perform hand seals or hand signs which allowed their chakra to manifest as techniques. As a kid, I used to mimick these hand signs and scream out the techniques associated with them but I never really knew any of the hand signs. I just randomly made them up.

One busy day after school back in 2018, I was playing games on my computer while searching for music to play in the background on Youtube. Kpop was big on my school back then, so I gave it a try. From then on, I was hooked and became a fan. Twice was famous among my peers so I chose them as my first group to “stan.” Watching one of their performances on Youtube, I encountered a problem that I’m sure many of those who are new to Kpop are very familiar with, they all look the same.