Scenarios¶

Scenarios form the building blocks of VisionAI platform. These scenarios are organized into Suites. Below we talk about different suites and the scenarios that are part of them.

All scenarios are available as pick-n-choose scenarios. You can pick the scenarios you want based on your business needs. Each scenario is independently tested.
Events provided by these scenarios are given below. Events are sent to Redis & Azure EventHub pubsub systems for further integration.
There are a few common events supported by all scenarios (daily summary, weekly summary etc.)
Currently supported scenarios are highlighted by a ✅. Roadmap scenarios are highlighted by a 📅.
Each of the scenarios can be quickly tested through visionai run <scenario-name> command. For example:

visionai run smoke-and-fire-detection

New scenario request

This section lists down all the scenarios that are supported by the VisionAI platform. There are more scenarios added daily - please send a request to us about any additional scenarios you need.

Privacy Suite¶

For a majority of organizations - employee privacy is a top concern. Along with employee privacy, the organization needs to make sure that any data does not leave the premises. Any faces detected through Vision AI system need to be blurred, along with text, signage, computer screens and other sensitive information.

Before any other scenarios are run, or before we store or process the images - the images are pre-processed through this privacy suite. As such, privacy suite is treated differently from other scenarios. Below examples provide a high-level overview of the privacy suite.

Status	Scenario name	Details	Additional considerations
✅	`face-blurring`	Blur any faces detected	More details
✅	`text-blurring`	Blue any text detected (paper, computer screens etc)	More details
✅	`license-plate-blurring`	Blur any license plates detected	More details
📅	`signs-blurring`	Blur any signs detected	More details
📅	`obstructed-camera`	If camera feed is obstructed, send an alert	More details

Hazard Warnings Suite¶

Following scenarios provide hazard warning examples supported by VisionAI suite. Currently supported scenarios are highlighted by a ✅.

Status	Scenario name	Supported Events	Additional considerations
✅	`smoke-and-fire-detection`	`Smoke event detected` `Fire event detected` `Sparks detected` `Open flames detection`	More details
✅	`no-smoking-zone`	`Smoking event detected` `Vaping event detected`	More details
📅	`spills-and-leak-detection`	`Water puddle detected` `Water leak from equipment detected` `Spill event detected` `Slippery sign detected`
📅	`missing-fire-extinguisher`	`Fire extinguisher missing`
📅	`blocked-exit-monitoring`	`Blocked exit detected`
✅	`rust-and-corrosion-detection`	`Rust or corrosion event detected`	More details

Worker Health & Safety Suite¶

Following scenarios provide Worker Health and Safety examples supported by VisionAI suite. (Also referred to as Personnel Health and Safety).

Workplace Personnel Health & Safety is important because it ensures that employees are safe and healthy in their work environment. This includes providing a safe and healthy work environment, proper safety training, and regular safety inspections. Additionally, it also includes enforcing safety policies to ensure that all employees are aware of and follow safety procedures, as well as encouraging a culture of safety within the workplace.

Currently supported scenarios are highlighted by a ✅.

You can see real-time events generated as soon as person is detected without PPE (helmets, gloves, safety boots etc.). There are options to configure what PPE's are required for your scenario. This can be done through the VisionAI web-application which can be accessed on through http://localhost:3001.

Status	Scenario name	Supported Events	Additional considerations
✅	`ppe-detection`	`Person detected without helmet` `Person detected without gloves` `Person detected without safety boots` `Person detected without safety goggles` `Person detected without face mask` `Person detected without vest` `Person detected without full-body suit` `Person detected without PFAS` `Person detected without ear protection`	More details
✅	`working-at-heights`	`Person detected without PFAS` `Steps detected without railings` `Person detected at height without parapets` `Ladder detected not in compliance`	More details
✅	`fall-and-accident-detection`	`Person slip & fall detected` `Potential collision/accident detected` `Wet floor detected` `Debris detected on floor` `Wet/slippery sign detected`
✅	`posture-and-ergonomics`	`Bend count per individual`	Straight camera angle More details
📅	`empty-pallets-detection`	`Empty pallets detected` `Partially empty pallets detected`
📅	`spills-and-leaks-detection`	`Water puddle detected` `Water leak from equipment detected` `Wet floor detected` `Spill event detected` `Slippery sign detected`
📅	`hand-wash-compliance`	`Missed hand wash`
✅	`confined-spaces-monitoring`	`Person detected` `Person left` `Person dwell time exceeds limit` `Person detected without motion` `Person fall detected`	More details

Occupancy Policies¶

Occupancy Policies relate to counting and tracking employees and/or other personnel in the room. These could include people-counting and enforcing max-occupancy policies, or tracking people's dwell time in a confined space.

Currently supported scenarios are highlighted by a ✅.

Occupancy Metrics

Occupancy metrics is similar in structure to max-occupancy, or restricted areas scenarios.
However it sends out a summary event is structured like this. This will give a granular summary event at the end of the day.

Users can start with occupancy-metrics and then move to max-occupancy or restricted areas if they need to enforce policies.

{
  "date": "2023-02-23",
  "stations": [{
    "id": "station_1",
    "hours": [
        {
          "start_time": "2023-02-23T14:00:01",
          "end_time": "2023-02-23T15:00:00",
          "occupancy_cnt": 14
        }
        ...
    ]
  }...]
}

Also need to specify that the camera needs to be configured to have a good view of the stations where occupancy metrics need to be checked.

Status	Scenario name	Supported Events	Additional considerations
✅	`max-occupancy`	`Person count exceeds max limit`	More details
✅	`restricted-areas`	`Person detected in restricted area` `Movement detected in restricted area` `Person detected after hours` `Movement detected after hours`	More details
✅	`dwell-time`	`Person detected` `Person left` `Person dwell time exceeds limit` `Person detected without motion` `Person fall detected`	More details
✅	`station-occupancy`	`Daily summary event`	More details
📅	`occupancy-metrics`	`Daily summary event`
📅	`authorized-personnel-only`	`Unauthorized person detected`

Company Policies¶

Company policies include specific scenarios that are relevant to your company. These could include scenarios like no-smoking/no-vaping zones, no food or drinks in certain areas, or no cell phones/pictures in certain areas. Some of these scenarios overlap with occupancy policies, but they are still useful to have here as separate scenarios.

Status	Scenario name	Supported Events	Additional considerations
📅	`no-food-or-drinks-allowed`	`Person with food detected` `Person with drinks detected` `Spill event detected`	More details
✅	`no-phone-text-pictures`	`Cellphone usage detected` `Person detected taking pictures`	More details
✅	`no-smoking-or-vaping`	`Smoking event detected` `Vaping event detected`	More details
✅	`no-children-pets-visitors`	`Children detected` `Pets detected` `Visitors detected`	More details

Suspicious Activity detection¶

Suspicious activity detection suite relies on a combination of activity detection models and object detection models. These models are trained to detect suspicious activity in a variety of scenarios.

Status	Scenario name	Supported Events	Additional considerations
✅	`vandalism-graffiti-company-property-destruction`	`Motion detected in area (gross event)` `People detected in area (more granular event)` `Non-uniformed personnel detected in area` `Non badged personnel detected in area` `Vandalism detected in area (before & after)` `Paint/graffiti detected in area (before & after changes)` `Behavior analysis event showing company property destruction.`	More details
✅	`firearms-knives-detection`	`Person brandishing firearm` `Person brandishing knives`	More details

Next Steps¶

Now that you have a better understanding of the scenarios that are available, you can start to think about how you can organize these scenarios into a solution that meets your needs. You can also go to the individual scenario page to learn more about it. We can customize each of these models for your use-cases and provide you with a solution that is tailored to your needs. You can contact us through this page.

This works by detecting a person's uniform and comparing it to a list of authorized personnel. This is a more advanced scenario and requires a custom model to be trained for your specific use-case. ↩