Monitor camera system

Abstract

To solve a problem in which, previously, video footage from monitor cameras has been simply recorded, and when something happened, a manual effort was required to review the video footage after the fact (obtaining evidence), and however, this has not been responsive and the process of reviewing the video footage has been extremely difficult, and furthermore, when the entire video footage from monitor cameras has been recorded, it would have to be discarded (updated) early due to capacity issues.SOLUTION: A video footage from a monitor camera is sent via internet to the "AI and Operator Center," where it is recorded in four modes as shown in Figure 1. In the event of a major incident, the AI sends the recorded footage to the operator, who then sends it to the police if necessary.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] This technology involves adding automatic evaluation by AI to conventional surveillance camera systems, with the AI ​​sending videos of possible incidents or accidents to an operator, who then, at their discretion, notifies the police if necessary. [Background technology]

[0002] In Sawaguchi Yasuko's TV show "Kasouken no Onna," there are many scenes where everyone checks video recordings together, which is fine if you just watch from the sidelines, but if you actually do it yourself, you know that if you miss something, it's a big problem (a big responsibility), and it's a very difficult task to keep watching mediocre footage for hours on end. Furthermore, it is basically a method of confirming evidence after an incident or accident has occurred, and does not lead to the prevention of crimes. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2023-043795 Summary of the Invention [Problem to be solved by the invention]

[0004] In the past, surveillance camera footage was simply recorded, and if something happened, a manual process was used to check the video footage after the fact (obtaining evidence footage). However, this method was not swift and the process of checking the video footage was very difficult. Also, if the entire video footage from the surveillance camera was recorded, it would take up a lot of space. Due to problems, it had to be discarded (updated) early. [Means for solving the problem]

[0005] Video footage from the surveillance cameras is sent via the Internet to the AI ​​& Operator Center, where it is recorded in one of four modes, as shown in Figure 1. In other words, the AI ​​will record the video footage sent from the surveillance camera in its original form, as well as for "minor incidents" (= guideline: 80% of all incidents: assumed to occur about once a day), "medium incidents" (= guideline: 17% of all incidents: assumed to occur about once a week), and "major incidents" (= guideline: about 3% of all incidents: assumed to occur about once a month), and will contact an operator when a major incident occurs, and the operator will then notify the police if necessary. In addition, original footage from surveillance cameras with large capacity will be discarded early, and the storage time for recordings will be extended in the order of recordings of minor anomalies → recordings of medium anomalies → recordings of major anomalies. [Effects of the Invention]

[0006] Since surveillance camera footage is checked in real time, signs of an incident or accident occurring can be reported immediately after it occurs, making it possible to prevent incidents and accidents from occurring or to resolve them quickly. If you need to check video recordings for some reason, you can simply check the recordings of major incidents, then medium-sized incidents, then minor incidents, making the task of checking video recordings easier. [Brief explanation of the drawings]

[0007] [Figure 1] This shows the gist of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0008] Video footage from the surveillance cameras is sent via the Internet to the AI ​​& Operator Center, where it is recorded in one of four modes, as shown in Figure 1. In other words, the AI ​​will record the video footage sent from the surveillance camera in its original form, as well as for "minor incidents" (= guideline: 80% of all incidents: assumed to occur about once a day), "medium incidents" (= guideline: 17% of all incidents: assumed to occur about once a week), and "major incidents" (= guideline: about 3% of all incidents: assumed to occur about once a month), and will contact an operator when a major incident occurs, and the operator will then notify the police if necessary.

[0009] Video footage from surveillance cameras is transmitted via the internet to the AI ​​& Operator Center. The data is sent to the AI ​​& Operator Center, where it is recorded in four modes, as shown in Figure 1. In other words, the AI ​​records the video footage sent from the surveillance camera in its original form, as well as for "minor anomalies" (= guideline: 80% of all anomalies; assumed to occur about once a day), "medium anomalies" (= guideline: 17% of all anomalies; assumed to occur about once a week), and "major anomalies" (= guideline: about 3% of all anomalies; assumed to occur about once a month). Examples of minor abnormalities include, "Judging from the overall situation, there is an ill-matched couple," "There is a stalker-like person following someone," and "There is someone who has just been injured." There are people who are (untreated)... Examples of unusual behavior include "someone taking pictures of a woman in a skirt from behind with their smartphone," "someone trying to forcefully take away a child or woman who is unwilling," and "someone riding a skateboard at a high speed, causing a dangerous situation." Examples of major incidents include "There is a fistfight," "There is a person carrying a dangerous knife," and "There is a person who is drunk and staggering while walking." Furthermore, even if something is initially thought to be a minor or medium-level anomaly, if it becomes statistically clear that it could lead to an actual incident or accident (= once the AI ​​has learned), it will be changed to a major anomaly, and the information will be communicated quickly from the AI ​​to the operator and then to the police. In the event of a major incident, the AI ​​will send partial footage of the actual incident or accident to an operator, who will then review the footage and, if necessary, send it to the police or fire department. Since the entire recording of the original video quickly consumes memory space, the original video recording is discarded early, and only partial video considered necessary is stored for a long period in a surveillance camera system, thereby saving memory space and leaving only the essential parts, making later analysis easier. [Industrial Applicability]

[0010] Since surveillance camera footage is checked in real time, signs of an incident or accident occurring can be reported immediately after it occurs, making it possible to prevent incidents and accidents from occurring or to resolve them quickly. If you need to check video recordings for some reason, you can simply check the recordings of major incidents, then medium-sized incidents, then minor incidents, making the task of checking video recordings easier.

Claims

[Claim 1] This surveillance camera system sends video footage from surveillance cameras via the internet to an AI & operator center, where AI records in multiple stages everything from partial footage of small signs that could lead to an incident to important partial footage of actual incidents or accidents. The AI ​​sends the partial footage of actual incidents or accidents to an operator, who then reviews the partial footage and, if necessary, sends it to the police or fire department. This surveillance camera system also allows for the early disposal of original video recordings and the long-term storage of only the partial footage that is deemed necessary.

Images