AI Layers are available to view in MapBrowser with a Nearmap subscription. You can visualize AI data raster layers covered by your subscription in your MapBrowser workspace as a default, and optionally overlay the layers and imagery with property or cadastre boundaries.
NOTE: To view raster AI Layers you need a subscription for Nearmap AI, and the required AI Packs. To view vector layers, you will also need a subscription for Nearmap Advanced AI Viewer. Contact your Nearmap Account manager to enable the required subscriptions.
Viewing AI layers
You can visualize AI Layers in MapBrowser to help with both data validation and to perform a visual search. Simply select the data layers you want to view in the Data Layers panel in the left to view them on the map.
NOTE: Once you select the AI layers, they will appear on the map unless you are not zoomed in enough - make sure you are at zoom level 16 or higher.
You can also retrieve all the data behind the vector visualization of AI Layers using the AI Feature API for integration in your own applications or in other geospatial tooling.
Selecting AI layer types
You can choose to view AI Layers as both vector and raster if you have a subscription to the Nearmap Advanced AI Viewer. Otherwise, only the raster AI Layers can be visualized.
NOTE: Some differences will be present as raster AI Layers are processed before various geospatial algorithms are used to create the vector AI Layers.
Without the Advanced AI Viewer subscription, vector layers will not be available. Your Nearmap account administrator can enable the Nearmap Advanced AI Viewer subscription and access for you in MyAccount.
To select a layer type:
- Click AI Layers in the Data Layers list.
- Click Settings. This option will be visible only if you have a subscription for the Nearmap Advanced AI Viewer.
- Select whether you want to view the layers as vector or raster. The selection will persist until you change it.
NOTE: Attributes and the selected display color are displayed only for vector layers.
Display and filter settings
To view the display and filter setting for a layer, click to select the layer in the Data Layer panel. The layer's display and filter settings are shown in the Inspector Panel along with the layer definition.
Use the slider to adjust the following settings:
- Colour - Each layer has a default display color assigned to it. To change the
default color setting, click the required color in the palette to select it.
- Opacity - This is the transparency applied to the layer on the map. Higher opacity reduces the transparency.
- Stroke Thickness - This is the width of the outline of the AI layer on the map.
- Background - This is the background setting for the map. A higher background value
causes the background to become darker. This setting is applied to all layers you view subsequently even if a different setting was applied for the layer earlier.
You can filter the AI data you want to view on the map by filtering the results based on the Confidence Score or the Fidelity Score depending upon the layer. These filters are a particularly helpful way to quickly choose thresholds for the data you want to include when using the AI Feature API. For example, is it more important to include all potential examples of a layer such as a building, or more important to ensure no incorrect examples are present?
Set the minimum and maximum thresholds for the acceptable score values based on which AI data will be filtered. Filtered data will be highlighted while the data outside the selected range will be greyed out. Commonly useful thresholds might be ignoring buildings below 70% confidence or 30% fidelity.
The Confidence Score expresses the likelihood of the object being a true example of say a building, shown as a percentage. The Fidelity Score is applicable only to buildings, and measures the quality of the shape of the vectorised building footprint polygon, shown as a number between 0 and 1.
NOTE: The minimum Confidence Score is different for vector and raster layers. For all vector layers except Poles, it is 50. For Poles, it is 30. For raster layers, a Confidence Score between 0-100 is applied to each point, and the lower the score is, the lighter the point.
Viewing AI layer attributes
To view AI layer attributes, click on the object (for example, building, pole, vegetation, etc). Details of the layer are displayed in a pop-up box. The details displayed depend upon the layer you have selected and your filter settings. The selected layer is highlighted.
For example, in the image above, we have selected a building for which the following attributes are displayed:
- Layer name
- Fidelity score
- Confidence score (in %)
- AI system version
- Building area
- Number of stories
NOTE: If you are viewing multiple layers, and the layers overlap, attribute information is displayed for the layer which is the largest in size when you click a layer.
NOTE: The relative transparency of raster AI layers reflects the confidence level of each pixel. A higher transparency reflects lower confidence in the correctly representation of an AI layer. The example below shows gable-shaped roofs with the more confidently identified roof above and the less confidently identified one below. For more information about Confidence, see Confidence Score.
Alpha and Beta layers
You may see an Alpha or Beta tag next to a layer name.
- Alpha layers are typically experimental in nature and may not have been tested against a specific use case, and may change substantially in meaning or be removed entirely in future releases of Nearmap AI.
- Beta layers are generally more stable, but may still change or be removed in future versions.
The best way to ensure the layers evolve in a way that meets your use case is to provide active feedback to the Alpha and Beta program.
❗️IMPORTANT: Alpha and Beta layers are only available as vectors.
Reporting AI data issues
If you notice any issue with an AI data layer you are viewing, follow these instructions to report the issue:
- Right-click on the layer and select Report AI data issue from the context menu. A form is displayed.
- Complete the form with the issue details including the layer name, issue type and details.
- Click SUBMIT.