"Face recognition is no longer a new identification technology. On the contrary, it is the oldest and most widely used form of identity recognition in human society. In recent years, it has seen a surge in practical applications, from airports and high-speed rail stations to banks, schools, convenience stores, and even concert venues. I still vividly remember how the 'crime-fighting hero' Jacky Cheung successfully completed his concert last year, living up to his reputation as the 'wind and rain bring prosperity, national peace and stability' in the entertainment industry. However, the credit for this achievement should mainly be attributed to face recognition technology.
Some criminals may think that they can attend Jacky Cheung's concert without leaving any fingerprints or revealing their identity by covering their faces. But they are mistaken, as the Russian Investigative Committee has developed a new 'recognition' technology that can identify a person's identity even when they are wearing a mask. This technology is a lethal weapon against 'masked criminals' and 'masked rioters'.
So how does face recognition work when someone is wearing a mask? In simple terms, it utilizes 3D facial modeling technology to determine the identity of a masked individual. Traditional face recognition relies on the three major features of the eyes, nose, and mouth. More advanced methods can also analyze the facial contour and the shape of the ears from a side view. For example, the size, position, and proportion of organs within a specific triangular area between the eyebrows and upper jaw are unique and invariant, making them suitable for identity recognition.
Even if a criminal is wearing a mask, their identity can be recognized as long as these facial features are exposed. When combined with a facial database, AI can automatically compare the features with the database to determine the identity of the masked individual. In fact, similar technologies have been developed as early as 2017 by students from the University of Cambridge, the Indian Academy of Sciences, and the National Institute of Technology in Vallangad.
These technologies use optical flow techniques to identify a person's face even when it is partially covered. By analyzing 14 points on the face, including 10 around the eyes, 1 on the nose, and 3 around the mouth, the technology calculates the distances and angles between these points to recognize a person's face. This technology provides a new means of capturing suspects and terrorists.
But what if the face is not captured at all? How can we identify the identity of 'masked rioters' and 'masked criminals'? This is where 'gait recognition' comes into play. Gait recognition is an AI technology that does not rely on facial features.
Did you know about the humorous 'Anatomy Prize' awarded in 2012? Two scientists from the Netherlands and the United States accidentally discovered that chimpanzees can recognize their companions based solely on pictures of their buttocks. Although it may sound like a joke, it is based on factual evidence.
However, humans find it difficult to recognize familiar individuals based on their buttocks. Instead, humans have developed other abilities to identify others, such as recognizing their walking posture and body shape.
The gait recognition technology has been tested for over 1000 hours in the public security system, participating in the investigation of more than 20 cases and retrieving 2000GB of public security videos. The gait recognition algorithm does not require the face to be visible. It identifies the target's identity by detecting their body posture and walking style. Even if a person is wearing a mask and walking casually with their back to a regular surveillance camera from several tens of meters away, the gait recognition algorithm can still determine their identity.
Furthermore, gait recognition can also calculate the density of a large crowd and perform real-time counting of a 1000-square-meter area with 1000 people located 100 meters away.
Compared to other AI recognition technologies, long-distance recognition is the core advantage of gait recognition. Iris recognition requires a distance of less than 30 centimeters, while face recognition requires a distance of less than 5 meters.
How does gait recognition work? Surveillance cameras capture a person's gait, and through detection and tracking, a video sequence of the gait is obtained. Key preprocessing and analysis are performed to extract the features of the person's walking, including motion detection, motion segmentation, and feature extraction.
After further processing, the extracted features are compared with the stored gait patterns in the database. If there is a match (or no match), an alarm is triggered.
With the advancement of technology and the support of artificial intelligence, the field of security seems to have gained the eyes of a god. In the eyes of this god, humans are all naked, and in the future, masked rioters and masked criminals will have nowhere to hide."
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.