Industreweb Visual Resource Monitoring Tool
The Visual Detection and Alerting system uses a Monitoring Box component running in the business premises or manufacturing facility to monitor using a camera and to recognise objects within its field of vision. It uses an Intel Visual Processing AI Unit to detect objects that it recognises from a pre-learnt AI model.
IW Collect running on the monitoring box then detects these events and based on a set of rules determines what Actions to perform. This could be to notify by email or SMS, to sound a siren, to light a warning lamp, push data to the EFPF cloud or display a message on a screen or dashboard.
Configuration The learning stage can take 1-5 days depending on the complexity and size of the visual library, and must be undertake as a setup step using an high powered PC provided by the commissioning company Control 2K. Preconfigured AI models supplied with the solution can be used to make implementations easier. Subsequently using the IW Global Administration tools the Events and Actions that should happen when an object is detected need to be created and deployed to the IW Collect node.
Operation Three scenarios are proposed for this technology in EFPF: • Detection and alerting of users not wearing mandatory PPE • User in an area where they should not be or where there is a health and safety risk they should be aware of (e.g. near a running machine or heavy loads are being moved) • Detection of staff not wearing mandatory COVID-19 protection When the system is executed it loads in its AI model and Actions and Events and starts it continual scanning. When an object is detected a probability of match is made, which if greater that a redefined user threshold the associated Actions can be carried out as shown in Figure 39.
Figure 39: Object detection example using the AI framework
The technology can be applied to any scenario where the system should detect a specific object or objects and trigger actions whether preventative, warning or confirmative Actions need to be undertaken.
Installation For technical details on installation, please refer to the Industreweb Wiki.