Smart factory: How to build an AI video supervision and risk early warning system for factory safety production?

1. Background requirements

Modern factories are mostly mechanically operated, guarded by a small number of people, or even unmanned management. Enterprises will place cameras in key areas such as production workshops and warehouses to store images as a traceability basis. However, tracing back to the source of the problem may have caused serious life safety accidents and property losses. Therefore, it is particularly important to enable the online monitoring video early warning system for each area of ​​the factory. It can not only fully grasp the real-time production situation of the enterprise, discover and deal with various abnormal problems in real time, but also supervise illegal operations and reduce personnel turnover. Inspection and supervision work to improve management efficiency.

Two, the solution

Through the video capabilities of the EasyCVR platform, the surveillance cameras in various areas of the scattered factories are unified and managed, and video surveillance live broadcast, cloud recording, cloud storage, video retrieval and review, intelligent alarm, platform cascading, and PTZ control are provided. And other capability services, combined with the AI ​​algorithm model deployed in the intelligent analysis gateway, can realize face detection, face recognition, pyrotechnics recognition, dangerous area intrusion detection, safety helmet/work clothes detection, off-duty sleep detection, equipment operation status detection Etc., constantly supervise and give early warning of unsafe behaviors and hidden dangers of factory personnel.

3. Technical Features

1. Video capability

In factory supervision, the video capabilities that EasyCVR can realize include:

1) Multi-protocol and device access

Cameras or hard disk video recorders in each area are connected to the EasyCVR platform in the general monitoring room through GB28181, Ehome, RTSP, HIKSDK, Dahua SDK and other protocols to achieve video aggregation.

2) Multi-terminal output

EasyCVR can distribute multiple protocol video streams, such as: WS-FLV, HTTP-FLV, HLS, WebRTC, RTSP, RTMP, can be integrated into mobile APP, applet, and other business platforms to play, and provide sharing links and iframe addresses, It can be played directly in the browser, which is very convenient for the supervision of security personnel.

3) Abnormal alarm

EasyCVR supports the device alarm reporting of the GB28181 protocol. Once an event is triggered, snapshots and videos will be captured immediately. If the monitoring personnel find anomalies, they can quickly locate the area to which they belong, which is conducive to timely processing and reducing the occurrence or expansion of accidents.

4) Video recording and playback

EasyCVR supports cloud recording (that is, configure the recording plan on the platform, store it locally on the server, and support cloud storage) and device recording (that is, front-end device recording, push to the EasyCVR platform through GB28181, Ehome, HIKSDK, Dahua SDK protocol, etc.), and can The device video playback, download, etc. can be used as the basis for video traceability.

2. AI intelligent detection ability

1) Inspection of hard hat and work clothes

Automatically detect and identify whether people entering the work area are wearing safety helmets or work clothes. If it is detected that a person is not wearing a helmet/wearing work clothes, an alarm can be triggered immediately. At the same time, managers can query alarm records and alarm screenshots according to the time period.

2) Off-duty sleep detection

Identify the working status of employees in important positions. When it is detected that employees are sleeping or leaving their posts without authorization, an alarm will be triggered and recorded.

3) Dangerous area intrusion detection

Factory workshops generally have many restricted areas, important storage areas and other places, and unauthorized personnel are not allowed to enter. The AI ​​algorithm can realize real-time detection and identification of human body. When someone enters the monitoring range, an alarm will be issued immediately and a snapshot will be taken, and a message will be sent to the management personnel for reminder.

4) Pyrotechnic recognition and detection

Fire hazards in the factory area cannot be ignored. AI algorithms are used to detect and identify fireworks in real time. Once smoke and flames are detected, an alarm will be triggered immediately. Firefighting devices can also be linked to perform operations such as spraying and extinguishing fires to prevent production safety accidents caused by fire hazards. , to protect the property of the factory and the safety of personnel.

Guess you like

Origin blog.csdn.net/TsingSee/article/details/131726403