Death Wish

The magic was in YOU.

Knowledge interconnected.

I stumbled upon some of this drawings and I’ve compiled some of them. Apparently they originated here in Tumblr:

Artistic Style Transfer For Videos

Graphics research from the University of Freiburg Computer Vision group have adapted the artistic Style Transfer method for video / moving image to great effect:

In the past, manually re-drawing an image in a certain artistic style required a professional artist and a long time. Doing this for a video sequence single-handed was beyond imagination. Nowadays computers provide new possibilities. We present an approach that transfers the style from one image (for example, a painting) to a whole video sequence. We make use of recent advances in style transfer in still images and propose new initializations and loss functions applicable to videos. This allows us to generate consistent and stable stylized video sequences, even in cases with large motion and strong occlusion. We show that the proposed method clearly outperforms simpler baselines both qualitatively and quantitatively. 

What You See Is What You Get

Promo for augmented glasses product Meta provides convincing demonstration of their technology through perspective of eyewear:

Our promotional videos are shot from behind the lens. We want to make sure people watching the videos are 100% certain that what they see is what they get with the real thing. For this reason, we designed and built a rig. 

Microscope Uses Artificial Intelligence to Find Cancer Cells

A microscope, invented by a professor at the University of California, uses artificial intelligence in order to locate cancer cells more efficiently than ever before. The device uses photonic time stretch and deep learning to analyze 36 million images every second without damaging the blood samples. This new technique for identifying problematic cells is faster and more accurate than standard methods currently in practice.

Commonly, doctors will add biochemicals to blood samples in order to check for cells containing cancer. The biochemicals attach what scientists call “biological labels” to damaged cells, which enables instruments to both locate and identify differences. These tests have proven problematic, as the biochemicals used would often damage cells, making them unusable for future testing. Other techniques currently in practice do not label cells, but identify cancer cells based on physical characteristics that can oftentimes falsely identify regular cells as damaged.

The photonic time stretch microscope images cells without causing them harm and can identify over two dozen physical characteristics, including: biomass, granularity and size. This alone makes for much more accurate and effective identification of the correct cells and makes retesting an option that wasn’t always available in the past. These new tests invented at UCLA use a photonic time stretch microscope, a practice capable of imaging cells in blood samples very quickly. Couple that with a deep learning computer program that locates cancer cells correctly 95% of the time, and you have a much better chance of pinpointing cancer cells early on, allowing for quicker treatment to stop the spread. Deep learning is a popularly used artificial intelligence that works with complex algorithms to pull meaning from data, leading to better decision making.

The most recent studies lead by Barham Jalali, professor and Northrop-Grumman OptoelectronicsChair in electrical engineering; Claire Lifan Chen, a UCLA doctoral student; and Ata Mahjoubfar, a UCLA postdoctoral student were published in the journal, Nature Scientific Reports. Jalali invented photonic time stretch. This new technology can prove to be helpful in many different scientific applications. For now, the main focus is its ability to take pictures of blood cells with the help of flashing lasers. The lasers can be compared to the flash of a camera and occur in a matter of nanoseconds, something normal instrumentation would not be able to digitize. With the help of optics that boost clarity within images while at the same time slowing them down just enough to detect and capture at a rate of 36 million images each second, the new microscope can track information not possible in the past. The deep learning function then distinguishes the difference between healthy and cancer riddled white blood cells.

Mahjoubfar says each frame is slowed down in real time and then amplified on an optical level so the data can be rapidly digitized. This lets scientists perform fast cell imaging that artificial intelligence is then able to classify. Chen states the photonic time stretch approach allows researchers to identify rogue cells in a very short period of time even with low levels of illumination.