1- Deep Vision Analytics Engine
The inReality Deep Vision Analytics Engine comes with pluggable best-in-class detection AI models built-in. It also provides a user-friendly user interface for configuring the AI models, gates, zones and live preview.
There are two methods used to configure Zones and Gates. In this guide the ‘Web User Interface’ method is reviewed. The second option is to use the DVA Settings in the platform’s Device Management section. For that option, please see the ‘Deep Vision Analytics - Zone & Gate Configuration - Platform Settings Guide’.
2- Access the Web User Interface
The web user interface can be accessed locally as described below.
2.1- Edge App - Deep Vision Analytics UI
- Open the Edge app which is installed on the same device where the DVA Engine is installed.

2.2- HTML5 - Browser
- You can also access it via a browser through the local network
- URL: http://<Host_Device_Local_IP>:8080/
3- Web User Interface
3.1- Navigation Menu

- The Video Analytics Navigation Menu is shown below:

3.2- Video Sources
In the main menu click on ‘Video Sources’ to view the options reviewed below:
- Sensor Settings - Gesture Recognition
- Face Detection Settings - Enter/ Exit
- Body Detection Settings - Detection Zone
- Gesture Recognition
- Enter/ Exit
- Detection Zone

3.2.1- Sensor Settings
- Camera Type: USBCamera or IPCamera
- Camera Serial Number: Unique serial number of the USB camera
- IP Camera URL: RTSP URL of the IP camera

3.2.2- Face Detection Settings
- Enable: Enable/Disable Face Detection
- Face mask detection: Enable/disable face mask detection
- Minimum Face Alignment Score (0-100%): Face detection head pose threshold
- Re-identification Matching Score (0-100%): Face re-identification matching confidence
- Live Preview - Detailed Information: Display advanced information on the live preview
- Deep Learning Model Inference Level: Face detection Deep Learning inference resolution

3.2.3- Body Detection Settings
- Enable: Enable/disable Body detection
- Re-identification Matching Score: Body detection re-identification matching confidence
- Skeletal detection minimum size: Skeletal detection minimum size
- Deep Learning Model: Body Detection or Skeletal Pose Detection

- Heatmap: Enable/disable motion based hourly heatmap
- Noise Filtering: Enable/disable filtering the noise in body detection
- Deep Learning Model Inference Level: Body detection Deep Learning inference resolution

3.2.4- Gesture Recognition
- Enable: Enable/Disable Gesture Recognition

3.2.5- Enter / Exit
- Data Source Code: Traffic datasource should be selected and can be identified by the letter ‘G’. Eg: A01G02
- Zone Type: Zone type should be selected as ‘Line zone’.
- Up: To position Entry/Exit towards the upward position of the Line/Gate zone


3.2.6- Detection Zone
- Data Source Code: Presence datasource should be selected and can be identified by the letter ‘P’. Eg: A01P02
- Zone Type: Zone type should be selected as ‘Detection zone’.
- Allow Interaction: Detection type
- HAND: Detection of hand in the zone.
- BODY: Detection of torso in the zone.
- FEET: Detection of feet in the zone.

Note: For HAND and FEET as ‘Allow Interaction’, the Deep Learning Model should be Skeletal Pose Detection


3.2.7- Line / Detection Zone

- Step 1: Go to Video Sources
- Step 2: Select N/A Zone 1 from right-side menu

- Step 3: Select the Zone Type according to the Data Source Code displayed
- Detection zone for Presence data source code (eg: A01P02)
- Line Zone for Traffic data source code (eg: A01G02)

- Step 4: Adjust the zones by clicking and dragging the vertices. The zones can be repositioned by clicking and dragging the zones.

The zones can be removed by clicking the cross button which appears when the mouse pointer is hovered over the zone.
- Step 5: Click Save
3.2.8- Non-Detection Zone
- Non-Detection Zones are used to exclude Face/Body Detection from a specific region in the camera FOV.

- Step 1: Click the square icon on the right top corner of the camera preview.
- Step 2: Define the Non-Detection zone.

- Step3: Select the Target type
- FACE: to avoid Face detection
- POSE: to avoid Body detection
3.3- Pipelines
- There are two options available, as shown below:
- Live Preview: Play a live preview of the camera for zone configuration.
- Start Recording: Record a live event.

3.4- Heatmap
- A motion-based hourly heatmap image of the camera preview is generated daily. The movement is tracked by color code, varying from Red to Green. Red indicates region with maximum movement and Green indicates region with minimum movement. See below.

3.5- Settings
3.5.1 - Subscription Status
- Subscription Status displays the subscription status of the Deep Vision Analytics application - Plan Name, Remaining days, Quantity, Due date and Status.


3.5.2 - Manage Output
- Manage Output displays the following data fields as shown below:
- File Name
- Time stamp
- Video Resolution
- Playback
- Download - The file can be downloaded to the host device to the folder:
/opt/apps/.data/com.inreality.python-ava/capture/videos

