Industrial Safety
Construction Control Automation
Construction Control Automation
Software (neural network) designed to automatically detect violations of safety on the construction site
Trial operation
Trial operation of the Construction Control Automation neural network was carried out during the construction of the capital's multifunctional complex.
On the construction site, the neural network successfully fixed:
–such safety violations as the absence of protective helmets and vests among workers and specialists, with the subsequent identification of the identity of the offender
–the presence of small items of working clothes
–presence of various objects in the forbidden or danger zone
Information on violations detected was promptly transmitted via the Internet to the BITRIX 24 system.
On the construction site, the neural network successfully fixed:
–such safety violations as the absence of protective helmets and vests among workers and specialists, with the subsequent identification of the identity of the offender
–the presence of small items of working clothes
–presence of various objects in the forbidden or danger zone
Information on violations detected was promptly transmitted via the Internet to the BITRIX 24 system.
According to its specifications, the software
Construction Control Automation
is a standalone executable application that is able to operate around the clock.
Construction Control Automation
is a standalone executable application that is able to operate around the clock.
Education of the neural networks in identifying various types of violations is carried out on photo and video, with people who observes and violates safety.
The neural network analyzes and records the received data and provides the consumer with information in the form of a report with an attached photo image of the violation, the date and time of the violation, its classification.
For a more detailed analysis of the situation, the neural network performs territory marking on a video frame indicating the dangerous and interested areas. In this case, the classification and segmentation of objects are performed. Moreover, the software is able to transfer information about the violation to contractors.
For a more detailed analysis of the situation, the neural network performs territory marking on a video frame indicating the dangerous and interested areas. In this case, the classification and segmentation of objects are performed. Moreover, the software is able to transfer information about the violation to contractors.
Reliability
Reliable (stable) functioning of the neural network is ensured by the implementation by the developer of a set of organizational and technical measures:
Аutomatic backup
Automatic software updates
Аutomaticsystem recovery
The recovery time of the neural network after a failure caused by a power failure or other external factors (except for a fatal operating system failure) does not exceed 30 minutes, subject to the hardware and software operation conditions.
To work with the software Construction Control Automation only one specialist is required.
He must:
→ Maintain a database of violations;
→ Solve problems of installation and maintain the system software -the operating system;
→ Install various programs;
→ Back up the database.
He must:
→ Maintain a database of violations;
→ Solve problems of installation and maintain the system software -the operating system;
→ Install various programs;
→ Back up the database.
Identificaion method
To date, testing has been conducted on methods of identifying the violator using face identification and the number printed on the vest.
As shown by a comparative analysis, the number printed on the vest or helmet is suitable for determining the identity of the violator to the organization, which is sufficient to alert the violating organization.
To date, testing has been conducted on methods of identifying the violator using face identification and the number printed on the vest.
As shown by a comparative analysis, the number printed on the vest or helmet is suitable for determining the identity of the violator to the organization, which is sufficient to alert the violating organization.
Technical requirements
IP video surveillance systems with rotary mechanisms and zoom, as well as a matrix resolution of at least 4 megapixels.
System scalability support.
-------------------------------------------------
Recommended server hardware parameters
of one server for 8 cameras:
CPU: Intel Core i5 (7th Gen) or higher;
GPU: 4 NvidiaGeForce 1080 or higher video cards;
SSD: 512Gb and more;
Network: Ethernet 1000;
RAM: 16Gb.
System scalability support.
-------------------------------------------------
Recommended server hardware parameters
of one server for 8 cameras:
CPU: Intel Core i5 (7th Gen) or higher;
GPU: 4 NvidiaGeForce 1080 or higher video cards;
SSD: 512Gb and more;
Network: Ethernet 1000;
RAM: 16Gb.
Comparative requirements for the size of identification elements, according to the requirements of the European standard EN50 132-7 for the detection, recognition and identification of people
1) subject to tilting the face to the camera no more than 45°
2) for cameras, with a horizontal field of view of 25 °
3) for numbers, having a width of 20cm, printed on the vest
2) for cameras, with a horizontal field of view of 25 °
3) for numbers, having a width of 20cm, printed on the vest
Recommended parameters of surveillance cameras
Construction Control Automation
In addition to the construction industry, this software (neural network) can be used quite effectively in the oil and gas industry, transport and logistics infrastructure, medicine and emergency situations.
© 2020 BSVT– New Technologies LLC
APPLICANTS
hi@bsvt-nt.by
COOPERATION
+375 17 396 15 16
info@bsvt-nt.by
info@bsvt-nt.by
OUR ADDRESS
220070, Republic of Belarus, Minsk, Soltysa Str. 187, ABС, 3rd floor, room 20