You can freely create a virtual fence on the screen in a work area where people (workers) should not be present.
When installed and a person (worker, human body) is detected inside the virtual fence, an alarm signal* (siren, etc.) is issued.
is issued and linked to the facility control panel to immediately and automatically stop the facility to prevent industrial accidents (serious accidents).
This is a solution that establishes a safe work environment through proactive prevention.
The virtual safety fence can be created in a rectangular shape according to the user's workplace environment and process characteristics.
It can be set freely without being limited by the shape of the work space such as polyline.
Even if it is inside the work site through A.I learning, if it is outside the scope of the virtual fence,
Operator detects but no alarm sounds.
A worker within the virtual fence has been detected and an emergency alert has been issued (red screen window)
Normal state with no workers detected within the virtual fence (green screen window)
Red border in the screen window: Virtual safety fence (can be freely set regardless of field conditions)
Industrial safety A.I solution application areas
Robot automation line
Cargo lift lower part
Assembly and logistics workshop
welding workshop
Operating principle of industrial safety A.I solution
Basic principle – human body detection solution based on thermal imaging sensor
Detect human shape and motion with thermal imaging sensor
Connect the dots to designate a free virtual safety fence area
A warning signal is generated when a person enters the virtual safety fence.
Developed an object detection AI model based on the algorithm used to develop the AI model, You Only Look Once (YOLO).
An algorithm that has the highest speed and accuracy among object detection algorithms and can predict the size, location, and type of objects.
Labeling of extracted image data is performed by drawing a bounding box surrounding the object inside the image.
The OLO model detects similar parts during actual prediction based on the size information of the center coordinates of the relevant Bounding Box.
Deep Learning (YOLO v5): Sets the reliability of the person in the input image and extracts location information that exceeds that level. In simple terms, it can be seen as ‘detecting a person’.
CUDA (GPU Parallel Computation): Real-time object detection and simultaneous processing of multiple cameras through parallel computation of GPU cores to accelerate deep learning.
The location of the person extracted through deep learning and parallel processing is colored red if the person is in the danger zone, orange for a warning, and green if the person is not in the danger zone.
Detects changes in heat and generates a fire prevention alarm and signal when a certain temperature is detected above the area setting area.
Composition of industrial safety AI solutions
By developing an AI-specific model that recognizes people using an IR sensor, we set a risk area and transmit a signal to the equipment used to prevent safety accidents from dangerous elements.
A thermal imaging camera and a control controller are defined as the basic package, and batch monitoring and reverse designation are possible.
On the control PC, you can check thermal imaging sensor status information in the form of a dashboard, and if necessary, you can remotely access a specific thermal imaging sensor to control the screen in real time.
When a human entry signal is detected within the virtual safety fence, a warning light or warning sound sounds and can be linked to a PLC according to predefined contents.
When a human entry detection signal occurs, LOG information can be recorded and confirmed.
Which sensor generated the detection signal can be checked on the control PC. (Cannot be checked on the PLC)
Use the easy and convenient learning function for fire surveillance auxiliary functions and sites where human misidentification occurs.
Send/receive signals to and from PLC through control controller built-in I/O
We refuse unauthorized collection of e-mail addresses posted on this website using e-mail collection programs or other technical devices,
and please keep in mind that violations will result in criminal penalties under the Information and Communication Network Act
※ Act on Promotion of Information and Communications Network Utilization and Information Protection, etc.
* Article 50-2 (Prohibition of unauthorized collection of e-mail addresses, etc.)
No one shall collect e-mail addresses using a program or other technical device that automatically collects e-mail addresses from an Internet homepage where the intention to refuse collection of e-mail addresses is stated.
No one shall sell or distribute the e-mail addresses collected in violation of the provisions of Paragraph 1.
Anyone who knows that the collection, sale, and distribution of the e-mail address is prohibited in accordance with the provisions of paragraphs 1 and 2 shall not use it for information transmission.
* Article 74 (Penalty) A person who falls under any of the following subparagraphs shall be punished by a fine not exceeding 10 million won.
A person who displays, sells, or displays for sale in violation of Article 8 (4)
A person who distributes, sells, rents, or publicly displays obscene code, words, sound, image, or video in violation of Article 44-7 (1) 1
A person who repeatedly reaches the other party with code, words, sound, image, or video that induces fear or anxiety in violation of Article 44-7 (1) 3
A person who takes technical measures in violation of Article 50 (6)
A person who transmits advertising information in violation of Article 50-8 A person who collects, sells, distributes, or uses e-mail addresses for information transmission in violation of Article 50-2
A person who transmits advertising information in violation of Article 50-8
A person who fails to register changes in registered matters, transfer or acquisition of business, or report merger or inheritance in violation of Article 53 (4)
