Matt's Homepage

Daniel Coppen recently teamed up with media artist Tomo Kihara to develop “How (not) to get hit by a self-driving car,” a street-based game designed to improve people detection in autonomous vehicles by challenging people to avoid being recognized by an object-detection algorithm. 

Participants use creative maneuvers like cartwheels and disguises to test and potentially enhance the AI’s ability to identify pedestrians in varied and unpredictable scenarios.  The game’s creators hope to conduct a global tour to gather diverse data, aiming to share it with researchers and self-driving car developers for better training of these systems.

Article:

• https://bigthink.com/the-present/you-win-this-street-game-by-getting-hit-by-a-self-driving-car/

Old games often suffer from the limited graphics capabilities of the time they were made, while developing new games costs a fortune due to the requirements to author high quality models and textures. What if you could solve BOTH problems – with the same solution? A machine learning project from Intel Labs in 2021 called “Enhancing Photorealism Enhancement” might push rendering toward photorealism a lot quicker and easier.

Researchers studied how to use a convolution network to re-render the scene. Below you can see an example of how they used the CityScapes dataset to give a much more realistic output of a race game – all in realtime.

You can read how the image enhancement actually works in their paper (PDF). It includes a lot of good information about how their method works and how it improves on previous attempts that have issues with color, object hallucination, and temporal instability. They do this by using the extra information provided by rendered scenes such as clever use of the g-buffer – along with a specialized discriminator and segmentation network.

Sources:

• https://www.theverge.com/2021/5/12/22432945/intel-gta-v-realistic-machine-learning-cityscapes-dataset
• https://isl-org.github.io/PhotorealismEnhancement/