Last week’s look at AI and ML news includes: “Don’t you dare call me a cartman!”, USCG Icebreaker Captures Arctic Images For Computer Vision-Powered Research, AI Brings Night Vision Into Focus, and CV Researchers Use Motion to Discover Objects in Videos
“Don’t You Dare Call Me A Cartman!”: Generative AI Makes Users The Stars of South Park Episodes
Solely in R&D for now, Fable Simulation’s GenAI tools handle the animation, voices, and editing.
Fable Simulation, a US company, has unveiled a groundbreaking generative AI-based tool called AI Showrunner that can create original South Park episodes with users as the main characters. The tool goes beyond just generating dialogue; it handles animation, voices, and editing, making it an all-in-one AI-powered TV show creator. Users only need to enter a short prompt, and the tool generates an entire episode, complete with a character resembling the user and using their voice.
Fable Simulation is working on AI Showrunner strictly for research purposes and will not be made available to the public. The company created a South Park episode using the tool as an illustrative example of generative TV but clarified that they won’t profit from it or allow others to use it without permission.
The creators of South Park, Trey Parker and Matt Stone, along with Comedy Central, the show’s broadcaster, were not involved or consulted in the experiment. Interestingly, South Park itself released an episode written partially using ChatGPT, demonstrating the growing influence of AI in TV production.
The role of generative AI in TV and film has become a point of concern during the Hollywood strike, with writers and actors worried about potential job displacement by machines. Intellectual property rights have also become a contentious issue around AI models, prompting Fable Simulation to emphasize their commitment to work with IP holders before considering public release.
Although AI Showrunner won’t be available to the public, Fable Simulation is exploring potential collaborations with studios and IP holders to allow fans to create their own episodes with proper permissions. The idea of fans contributing creatively to shows through AI-generated content is seen as an interesting prospect.
For now, South Park Studios has not issued any comments regarding this AI experiment.
US Coast Guard Icebreaker Captures Arctic Images For Computer Vision-Powered Research
In an ambitious scientific venture, the U.S. Coast Guard (USCG) icebreaker Healy is embarking on a remarkable journey across the North Pole, aimed at capturing crucial images of the Arctic to support cutting-edge research on this rapidly changing region. To achieve this, researchers from MIT Lincoln Laboratory installed a state-of-the-art camera system on the Healy before its three-month science mission began on July 11. The resulting dataset will serve as a powerful resource for developing computer vision tools tailored to analyze Arctic imagery.
Valuable visual data will be used for computer vision-powered analysis for climate monitoring, wildlife tracking, and more. (Image generated with AI)
This initiative holds great promise for enhancing both the safety and efficiency of navigation for mariners while contributing significantly to maritime domain awareness. Moreover, the study of AI analysis in this unique environment will be invaluable for tackling the emerging national security challenges posed by increased traffic in the Arctic, involving military vessels and illegal fishing ships. Additionally, it will shed light on critical questions concerning the region’s changing climate, wildlife, and geography.
The Healy, the largest and most technologically advanced icebreaker in the USCG fleet, is an ideal candidate for collecting this dataset. Collaborating with the USCG Research and Development Center, the team developed the CRISP (Cold Region Imaging and Surveillance Platform) system, which boasts a long-wave infrared camera designed to withstand harsh maritime conditions. Capabilities include stabilization during rough seas and the ability to capture imagery in complete darkness, fog, and glare. It records both video and still images with GPS data.
Currently, the availability of imagery datasets for studying these transformations in the Arctic is limited. Satellite and aircraft images only offer a restricted scope of information. However, a camera mounted on a ship can provide more detailed and comprehensive images of the environment, capturing different angles and even other ships.
The resulting dataset is expected to be a substantial 4 terabytes in size and will be made publicly available once the USCG science mission concludes. By sharing this data with the wider research community, experts can develop enhanced tools for operating in the Arctic. To further aid researchers, the Lincoln Laboratory team plans to provide a baseline object-detection model and create classifiers to identify and track specific objects across images.
This innovative project reaches far beyond its immediate impact on USCG missions. It has the potential to advance AI applications aimed at combating climate change. Those interested will be able to download the dataset for free from the Lincoln Laboratory website, showcasing the lab’s commitment to harnessing AI to address national challenges—spanning public health crises to maritime awareness in the Arctic.
AI Brings Night Vision Into Focus and Innovation Breakthroughs for Autonomous Vehicles
A groundbreaking technique utilizing AI promises to revolutionize thermal imaging, offering clearer and more detailed images like those captured during daylight with conventional cameras.
HADAR powered by vision AI sharpens can potentially improve nighttime capabilities for self-driving cars.
Developed by researchers at Purdue University, the method, named heat-assisted detection and ranging (HADAR), aims to significantly improve night vision capabilities for various applications, particularly self-driving cars. By training a neural network to distinguish an object’s heat signature from environmental noise, vision AI powers sharp and precise thermal images, materials identification, and even offers crucial depth information.
Unlike radar and lidar systems, HADAR is entirely passive, making it ideal for a future with numerous autonomous vehicles where signal interference could pose risks. The outcomes could mean safer and more effective nighttime navigation for self-driving cars and other computer vision technologies.
The potential of thermal imaging to excel at nighttime is widely acknowledged by experts who also highlight its unique ability to determine object composition, making it valuable even in daylight when combined with traditional imaging methods.
There are challenges ahead. The team has only tested the method on still images, and they recognize the importance of improving the speed of data collection and handling motion blur. Despite this, researchers envision a broad array of applications for the technology, ranging from enhancing the safety of self-driving cars to aiding biologists in remotely tracking wildlife.
With ongoing R&D, the prospects for thermal imaging are promising, offering potential solutions to various industries and advancing our capabilities in both dark and well-lit environments.
Check out how Plainsight is leveraging thermal cameras to enable computer vision-powered tank fill-level monitoring, applicable across manufacturing and oil and gas sectors. Vision AI solutions empower remote measurement, tracking, and alerting for changes in tank content levels and protect workers, communities, and the environment.
Objects In Motion Stay In Motion: Computer Vision Researchers Use Motion to Discover Objects in Videos
Researchers from Carnegie Mellon University’s Robotics Institute are exploring how computer vision systems can detect objects in motion more effectively than stationary objects. The project was a collaboration between CMU, Toyota Research Institute, the University of California, Berkeley, and the University of Illinois Urbana-Champaign, with the sponsorship from Toyota Research Institute.
Computer vision enables object categorization rather than just specific instance recognition.
The main objective was to improve object recognition in real-world scenes, particularly for applications like autonomous driving, retail robotics, robotic manipulation, and home robots. The researchers developed a framework called MoTok, which allows the computer to identify features of moving objects and reconstruct them, facilitating the discovery of the same object again in different instances.
An interesting development was that the computer could visualize these features in a simplified manner, enabling it to categorize objects rather than just recognizing specific instances. Making computer vision more autonomous and scalable, this reduces the dependence on labeled data, which can be time-consuming and expensive to obtain.
Plainsight In the News
In a recent article in Vision Systems Design Plainsight’s work with MarineSitu was featured with a deep dive into our work together to develop advanced vision AI solutions for their breakthrough marine monitoring system.
MarineSitu’s Co-Founder and CEO, James Joslin, and Plainsight’s Co-Founder and CPO, Elizabeth Spears, discuss the ways computer vision is revolutionizing the renewable energy sector within the “blue economy”, providing insights into marine life and complex underwater environments, and ensuring sustainable marine resource development. Dive into the full article to discover all the details.
About the Author & Plainsight
Joan Silver is the SVP of Marketing at Plainsight and oversees the full scope of marketing and communications for the company. Propelling multiple companies from startup and launch, through funding rounds, to successful IPOs and acquisitions, Joan builds brands from big ideas to big time. A trailblazer in the B2B digital marketing industry, she’s a believer in the transformational power of AI and loves technology that solves problems and improves our daily lives.
Plainsight provides the unique combination of AI strategy, a vision AI platform, and deep learning expertise to develop, implement, and oversee transformative computer vision solutions for enterprises. Through the widest breadth of managed services and a vision AI platform for centralized processes and standardized pipelines, Plainsight makes computer vision repeatable and accountable across all enterprise vision AI initiatives. Plainsight solves problems where others have failed and empowers businesses across industries to realize the full potential of their visual data with the lowest barriers to production, fastest value generation, and monitoring for long-term success. For more information, visit plainsight.ai.