136894658
submission
shirappu writes:
AI technology is getting better and better at improving old video footage by adding new frames to improve smoothness and adding color to black and white footage (a great explainer for the techniques can be found here: https://youtu.be/EjVzjxihGvU). Now a Youtube channel is applying the same techniques to the moon landing in 1969.
The AI restored footage stabilizes the shaky old footage, adds frames, and motion smooths the whole thing to essentially bring the footage into the present. Though it can be difficult to improve these videos because high quality source footage is a necessity, it's still a good example of how improved these technologies are becoming with the development of AI support.
The actual improved video of the moon landing can be found here: https://youtu.be/mDYyWv5MiRg
135341731
submission
shirappu writes:
Daniel Voshart's work on creating life-life images of Roman emperors from their statues started as a quarantine project and quickly got out of hand. His portraits of the emperors (a collection of 54 as of July) are created using generative adversarial networks, which are fed images of the emperors from statues, coins, and paintings. These are then edited and tweaked based on historical descriptions, and reworked in photoshop, where Voshart says he can "avoid falling down the path into uncanny valley."
Voshart has written about the process himself here: https://medium.com/@voshart/ph...
134760849
submission
shirappu writes:
A study published on June 25th in IEEE Transactions on Multimedia looked at creating an automated program to analyze the movements in video frames and create sound effects to match the scene. In movies and television, this work is called Foley and is considered an important part of crafting the experience, but is also time consuming and sometimes costly.
The AI Foley system created in the study works by extracting image features from video frames to determine appropriate SFX. It then analyzes the action taking place in the video, and attempts to synthesize SFX to match what is happening in the video. In a survey where students were shown automated sound effects, 73% though the automated effects were more genuine than the original sound effects. It's worth noting that the automated system performed best when timing was less important (e.g. general weather) versus when timing was key (e.g. typing, thunderstorms).
The full report can be read here: https://ieeexplore.ieee.org/do...
133331736
submission
shirappu writes:
The development of AI systems for medical application is constantly ongoing, and ranges from chatbots and medical imaging to automating paperwork and predictive diagnosis. Many believe AI systems can ease the workload of doctors by predicting COVID-19 in CT scans and X-ray images.
However a group of medical statisticians is reviewing many of these predictive models as they are released, and so far they still aren't performing to a high enough level. The main problems appear to be (perhaps unsurprisingly) a lack of necessary data, and not enough domain expertise.
The leader of the above group of statisticians, Maarten van Smeden, pointed to a lack of cooperation between researchers as a road block in the way of developing truly accurate models. "You need expertise not only of the modeler," he said, "but you need statisticians, epidemiologists and clinicians to work together to make something that is actually useful."
133134718
submission
shirappu writes:
AI|Writer is an experiment in which artificial intelligence is used to simulate both real and fictitious famous personalities through written correspondence. Users can ask questions and receive explanations from simulated versions of Isaac Newton, Alfred Hitchcock, Marie Curie, Mary Shelley, and many more.
This application builds upon text generators such as GPT-2, which allowed for computer systems to generate believable (though often factually incorrect) text. So far these systems have been used to help write articles and even to create video games.
The creators of AI|Writer have used Open AI's API to build their system. This API allows the system to find patterns in text and complete them, which is how it was trained to respond in the voice of historical and popular characters.
However, the creators have made it very clear on their site that although the AI will respond in the voice and manner of a particular person, the AI response should not be consider a reliable source nor an accurate reflection of that person's views and opinions.
133076044
submission
shirappu writes:
Neural Synesthesia (link: https://vimeo.com/neuralsynest...) is an AI art project that creator Xander Steenbrugge calls a collaboration between man and machine. To create each piece, he feeds a generative network with curated image datasets and combines the ever-transforming results with music that is programmed to control the shifting visuals.
Steenbrugge is one of a growing number of artists using AI technology as a collaborative tool. In this interview he echoes what others have also said, that the strength of AI is that the artist is no longer limited by their own imagination, and that AI influence can result in unexpected and interesting ideas. He also comments on the growing trend of AI technology, and how it might lead to the creation of a new cinema 2.0-like experience.
132804006
submission
shirappu writes:
A new paper by Disney Research shows off a newly developed neural network that can swap faces in photos and videos at high-resolution. The idea behind this technology is to replace an actor's performance with a different actor's face, or for roles that require de-aging or increasing age, or for portraying actors who have passed away.
Current face-swapping technology (also known as deepfakes) often creates an "uncanny valley" effect, where something about the image or video feels off. Though there is some of that in Disney Research's tech, it's still a huge step forward for creating believable face-swapping in the entertainment industry.
This has once again brought up a conversation around the ethical use of this technology and the potential for malicious use (See: https://becominghuman.ai/a-loo...). However, given the amount of ongoing R&D in this area, it seems unlikely that we'll see any slowdown in the near future.
132266298
submission
shirappu writes:
The technology, called PIFuHD, takes photographs of people and reconstructs them in 3D. The tech works on deep neural networks with multi-level architecture that allows for high resolution and accuracy in 3D models even at low levels of memory. More is available in the detail-heavy research paper: https://arxiv.org/pdf/2004.004...
Applications for this kind of automated image digitization include medical imaging and virtual reality, and the researchers have released a version of the model for users to try out themselves on Google Colab (link: https://github.com/facebookres...)
132143962
submission
shirappu writes:
For eight years now, the Mars Rover Curiosity has been exploring the surface of Mars. Even now, it's still exploring, and still getting upgrades. According to Tech Crunch, NASA is now looking to interested volunteers to help upgrade the rover's terrain-scanning AI systems by annotating image data of the planet itself.
Volunteers go through a short tutorial after which they can label images to help the rover better understand the terrain on which it drives. The system is expected to be used in future planet rover robots, and the project marks an interesting example of open crowd-sourcing to improve machine learning systems, and how it is impacting technology even on other planets.
AI4Mars site link where people can volunteer: https://www.zooniverse.org/pro...
131983660
submission
shirappu writes:
Nick Walton created AI Dungeon as an experiment to build an AI dungeon master for D&D games. Since then, it's grown into a text adventure game in which players can type anything they want, with the game's AI responding dynamically (https://aidungeon.io/).
In this interview about the year since the release of the game, Nick talks about how it works, and the systems they're working on now to create quest systems, world persistence, and longer term memory. He also opens about where he thinks AI systems can support game development.
From the interview:
"One of our visions for AI is not as a tool to replace game designers, but a tool for augmenting their work. We want to make it easier to create awesome games. If it only takes one or two people to build an awesome game because AI fills in the details, it opens up doors for a lot of people."
131896474
submission
shirappu writes:
Thousands of senior citizens in South Korea are being monitored through virtual-assistant smart speaker technology. The speakers are designed to pick up on signs of danger as it pertains to loneliness and insecurity, and can recommend visits by public health officials. In essence, it is a tool to prevent elderly residents from dying alone, a growing issue in cities with aging populations.
However, as discussion and debate continues around the use of private data and how it is kept, questions are still being raised as to whether this data will be used for other industries or technology in the future.
