Portland, OR – September 19, 2023 – PreAct Technologies (PreAct), an Oregon-based developer of near-field flash LiDAR technology, says that future camera sensors will be able to determine the difference between real objects and AI-generated deep fakes. In the digital age, reality and fiction have started to blur into one another, often making it challenging to discern fact from fabrication. With the advent of deep fakes and advanced CGI, the fidelity of our visuals has been questioned repeatedly. Nowhere is this more evident than in computer vision. Computer vision has become a horizontal technology and is poised to significantly impact numerous industries, including automotive, medical imaging, retail, health and safety, security, and entertainment. Computer vision's social impact is only starting to come into focus (pun intended).
Deep Fakes and Their Dilemma
In the near term, computer vision will exacerbate problems with deep fakes in the media and communication industries as the authenticity of images and videos is under threat. A looming question resonates: How do we know if what we see is real? It's unsettling to imagine a scenario where we doubt whether the person we chat with on Zoom is who they claim to be. This brings forth the notion of dual-factor authentication for video calls — a step to ensure that the individual on the other side is genuine. In addition to this unease, companies are venturing into posthumous video chats, allowing individuals to converse with digital replicas of loved ones long after passing. The Washington Post aptly termed one such venture "the next step in the human quest for immortality." Moreover, the entertainment industry faces its own set of challenges. Hollywood is abuzz with concerns about AI potentially replacing actors and writers. This is not unfounded; recent films, like the latest Indiana Jones movie, showcased a de-aged Harrison Ford, presenting him in a rendition 40 years younger. The looming question arises: How far are we from movies being entirely created by AI, sans human involvement? This dystopian future does ring bells reminiscent of narratives from shows like "Black Mirror."
Combatting Fakes: The Role of Future Cameras
While the above might paint a grim picture, there's a silver lining. The very technology responsible for these concerns can also be the solution. As computer vision progresses, it can be employed to detect inconsistencies in AI-generated content.
Micro-Expressions and Biometrics: AI-generated faces, no matter how lifelike, might still struggle to perfectly replicate micro-expressions or minute facial movements that are inherently human. Cameras of the future, with high-resolution sensors and advanced algorithms, can be trained to identify these subtle differences.
Pixel Analysis: AI-generated visuals often leave behind patterns at the pixel level that can be indicative of their artificial nature. Future cameras can discern between factual and fabricated content by analyzing pixels for anomalies or inconsistencies.
Metadata Verification: Every digital image or video comes with metadata about when, where, and how it was captured. By verifying the integrity of this metadata, future cameras can validate the authenticity of the content.
Source Verification: Cameras can also be linked to Blockchain or similar decentralized databases to verify the source of the content, ensuring its legitimacy.
The Union of Depth Sensing and Blockchain: A Shield Against Digital Deception
While the advancement in AI and computer vision has paved the way for many innovative applications, it has also increased digitally manipulated content. This raises a concern: How do we discern reality from fabrication, especially in real-time? One proposed solution lies at the intersection of depth sensing and blockchain technology. Let's delve into how these two technologies can collaboratively combat the rise of deep fakes and digital forgeries.
Depth Sensing: More than Meets the Eye
Depth sensing, a technology used to measure the distance between the sensor and objects, adds an extra layer of information to an image or video. Unlike conventional 2D images, depth maps provide a 3D spatial understanding of the scene. This gives a unique signature to the captured moment, making it incredibly challenging for AI algorithms to generate depth data consistent with the real world.
Blockchain: A Digital Ledger of Authenticity
On the other hand, Blockchain is a decentralized ledger system renowned for its tamper-proof design. Once data is added to the Blockchain, it becomes immutable, meaning it cannot be altered without the network's consensus.
Real Time Timestamps: The Confluence of Depth and Decentralization
By combining depth-sensing with Blockchain, we can create real time timestamps that inherently distinguish genuine footage from forged content: Instantaneous Verification: As a video or image is captured, the Camera's depth sensor simultaneously records the depth map of the scene. This depth map and the traditional image or video data are immediately hashed (converted into a fixed-size string of numbers) and added to a blockchain.
Unique Signatures: The depth data is a unique signature, like a fingerprint. It's nearly impossible for AI-generated content to replicate the exact depth nuances of a real-world scene.
Immutable Timestamps: Once this data is recorded on the Blockchain, it creates an immutable timestamp that verifies when the content was captured. Any attempt to manipulate the image or video afterward would not match the original hash on the Blockchain, flagging it as potentially exploited.
Public Verification: Due to the decentralized nature of the Blockchain, anyone can validate the content against its timestamp, ensuring transparency and trustworthiness of the footage. Preact is building a new type of Camera. Here at PreAct, we are building the future of Sensor fusion and camera technology. Such a system has profound implications, especially for industries that rely heavily on the authenticity of visuals, such as industry, automotive, security, and law enforcement. For instance, news agencies can prove the integrity of their footage, and courts can validate the legitimacy of the evidence presented. However, it's worth noting that while this amalgamation of depth sensing and Blockchain presents a robust solution, it is possible.
Advanced AI algorithms will continue to evolve, and the arms race between digital deception and its detection will persist. Nevertheless, as we move into an increasingly digital future, the union of these technologies offers a promising shield against the onslaught of fabricated content, anchoring us to a reality that, while augmented, remains genuine at its core. While AI-generated content brings undeniable advantages, it also ushers in challenges that can undermine the authenticity of digital media. However, with advancements in computer vision, lidar technologies, and the capabilities of future cameras, we're gearing up for a reality where distinguishing between genuine and fabricated content will become increasingly feasible. The goal is not just to marvel at technology's prowess but to harness it responsibly for a truthful digital future.
About PreAct Technologies
PreAct Technologies is the market leader in near-field software-definable flash LiDAR technology and integrated SDK (software development kit). Its patent-pending suite of sensor technologies provides high resolution, affordable LiDAR solutions to a wide range of industries including robotics, healthcare, ITS, logistics, security, industrial, consumer electronics, trucking, and automotive. With unmatched quality and accuracy, PreAct’s edge processing algorithms drive technology resulting in 3D depth-maps of small objects at sub-centimeter accuracy up to 20 meters. PreAct’s LiDARs and SDK enable companies and innovators to address the industry’s most pressing business and technology needs. The firm is headquartered in Portland, Oregon, with offices in Ashburn, Virginia, and Barcelona Spain. For sales inquiries, please contact sales@preact-tech.com. For more information, visit www.preact-tech.com.
