2 minutes to locate, 30 minutes to retrieve a lost child
Driving while inhaling "laughing gas": New type of illegal act detected instantly
Speeding cars, faster enforcement – late-night disturbances hit the mute button
A few letters help solve a case, lost items quickly retrieved...
Recently, local public security authorities have utilized the enhanced HiKvision ViewLance large model.View Large Model Application Integrated MachineOur company's product line has launched AI features."Text-to-Image Search"Smart Applications Boost Emergency Response and Case Handling Efficiency
AI "Text-to-Image Search"Intelligent applications, leveraging multi-modal large model technology, enable cross-modal information retrieval between natural language and video images, through open textual semantic descriptions.People, vehicles, goodsRelevant characteristics,Enter "a sentence" or "a few words"The platform allows for quick searches for missing persons, lost items, and illegal activities, aiding in swift recovery efforts and law enforcement cases.
2 minutes to lock in the location
Lost Child Found Within Half an Hour
The golden time window for finding missing individuals is only 24 hours, yet the difficulty of the search is akin to finding a needle in a haystack.
Recently, a local police station received an urgent report from Mr. An, a citizen, who reported that his 10-year-old daughter went missing near the entrance of a store.
Upon receiving the alert, the criminal investigation team of the video surveillance section at the local Public Security Bureau, based on the characteristics of the child provided by the parents, entered the relevant information into the platform.Pink T-shirt, jeans, little girlBy searching with relevant keywords, we retrieved the "key clue image" of the missing little girl and then used the image to search for similar images within the system.In less than 2 minutes, we located the latest position of the young girl.Officers successfully located and retrieved him based on location information. The entire process from receiving the alarm to finding him took less than half an hour.
If searching for individuals through traditional methods, there are issues such as lower judgment efficiency and poorer timeliness. Sometimes, staff have to review image libraries containing tens of thousands of photos. When faced with massive video and image data, a significant amount of manpower is required for analysis, which can easily lead to missing clues due to visual fatigue. The entire analysis process often takes several hours or even an entire day, directly impacting the golden time window for locating missing individuals.
In cases of missing persons, a critical challenge in the investigation is identifying key clue images. If they can be found quickly, the subsequent process of tracking them down can proceed efficiently. The application of large models has greatly facilitated this aspect.Significantly boosted the efficiency of clues search for missing individualsThe local police department recently quickly located several missing individuals, all of whom were found through the rapid extraction of key images from the platform.
Driving while inhaling "laughing gas"
New Illegal Activity "Instant Search"
Driving while "blowing up" balloons? You read that right; the balloons in question are filled with a hazardous chemical—nitrous oxide. Inhaling nitrous oxide while driving can numb the nerves, obstruct vision, and pose significant traffic safety risks.
In response to such new illegal activities, police officers have a solution. This June, a local public security department utilized AI "text-to-image" search."Blowing up balloons while driving"An individual was found "blowing up balloons" while driving, suspected of having inhaled laughing gas. Based on precise analysis of leads and thorough investigation,The case was swiftly solved in just one day.The company has prevented the continuous occurrence of the dangerous practice of smoking laughing gas while driving.
Additionally, local public security authorities have been utilizing AI "text-to-image" search to investigate illegal sales of "laughing gas," effectively preventing the illegal dissemination of hazardous chemicals.
Speeding cars? We're on it faster!
Motorcycle "Speeding and Disturbing the Peace" Recognition Achieves 91% Accuracy
As night falls and the streets grow quiet, the occasional roar of engine notes echoes through the streets at over 100 decibels, equivalent to a pressure drill operating at full throttle. For nearby residents, it's an unbearable disturbance. Moreover, the vehicles involved in street racing often have illegally modified exhaust systems, which are highly prone to causing traffic safety accidents.
However, the illegal activity of street racing is mobile, with no fixed time or location, making it difficult to investigate. Traditionally, it is mainly discovered through public complaints and police officers reviewing video footage in their districts.
During this year's college entrance examination, in order to provide a quiet and comfortable preparation environment for all candidates in the city, the local public security department utilized AI's "text-to-image" search function to launch a special campaign against "high-speed driving and loud noise". The keywords entered were:Motorcycle StuntsThe AI large model identified corresponding targets, uncovering multiple high-speed driving and illegal street racing cases, spanning both day and night. The silent action during the college entrance examination yielded positive results.
Coincidentally, another regional public security department has adopted large model fine-tuning to enhance recognition accuracy, after a period of application,Search accuracy for "high-speed driving" and "motorcycle stunts" reaches 91%.Brings more flexible, efficient, and accurate technological means for investigating and cracking down on illegal street racing.
A Few Letters Help Solve the Crime
Facilitate Quick Return of Lost Items
A theft occurred under the dark night and strong winds, with only a few letters as clues... AI's "text-to-image" search technology swiftly cracked the case, aiding in the quicker recovery of the lost items.
Recently, residents in a local area reported a burglary in an underground parking garage early in the morning. However, the video captured by the garage's surveillance camera was extremely blurry, only barely showing the suspect wearing a white T-shirt with two lines of English letters, resembling "STAR**R."
Officers utilized AI "text-to-image" searchWhite, T-shirt, Male, Black Letters, STAR**RFollowing these keywords, the search results revealed that the clothing features of the man in the snapshot were highly similar to the suspect in the garage video. Based on the snapshot, police officers conducted multiple image condition overlays to search, ultimately confirming the suspect's identity and solving the theft case.
