Mapflow - Get started

Go to Mapflow and register or login using your email. You can also use Google account to login. Mapflow processes imagery and extracts vector objects from it. So as a user, you start processing runs, or as we would call it, AI-mapping.

Upon the registration, the Default project will be created automatically and you will be redirected to the main dashboard:

You can create as many projects as you want to organise your work and share the project access with your coworkers.

Note

See about the Mapflow Team accounts to collaborate with other users. 👈

When starting a new project, you can set a name, customize icons, and modify a list of available AI-mapping workflows that are enabled in your project.

Note

If your account is upgraded to the Premium or Custom – you might be able to customize your project mapping workflows and some of their options.

Glossary of terms

• Processing – a single execution of an AI model on a specific geographic area

• Workflow/Flow – the AI model configuration pipeline definition

• Project – space for multiple processing runs

Mapflow My account

This section contains tabs for user account settings and services.

The following services are available here:

1. User info

• Fill in the user profile details

• Set up your preferred language from the available

• Enable email notifications (about the processing status, etc.)

2. Balance

Top up your Mapflow account with Mapflow credits.

3. Billing history

Here you can see your processing history with the details. If you are a Team account owner, you can search through the history of the entire team processing runs. Use filters to sort by date and processing status. To filter by the custom parameters like user email or processing name - start typing in the search input.

Note

You can export billing history as CSV.

4. Team

Create and manage your Team accounts.

5. My Imagery

Manage your imagery collections that you can currently use in the Mapflow QGIS plugin.

6. API

Obtain or reset your Mapflow API token.

How to run your AI mapping workflow in Web app

Mapflow is designed to be intuitive. Here is our step-by-step user guide:

1. Create a project

You can use the Default project created upon registration or create a new one. To create a new project, click the “Create project” button and fill in the project name and description.

Note

You can manage a set of models linked to a project (click the button on the project card):

2. Specify the AOI (Area Of Interest)

First step is to set the area of analysis for your project: by AOI polygon or by the extent of your GeoTIFF image.

2.1. Select AOI

The user can draw the area using Draw rectangle / Draw polygon tool or upload it in GeoJSON format (draw and upload, as well as view the data structure, follow this link - geojson.io).

Warning

Note that max limit of total area is 25 km² for the free plan. The area limit is calculated not by the polygon, but by the Bounding Box (Bbox). Therefore, in this example a warning MAX BBOX 25 KM² appears in the example below (bounding box is highlighted):

If you want to extend your AOI limit, switch to the paid plan.

2.2 Upload GeoTIFF

You can upload your aerial images in GeoTiff format. After the image is uploaded, you need to draw/upload an AOI or click “Use Image Extent”.

Warning

Currently, a preview of the uploaded image is not possible before the processing, you will see only the area of its extent.

The processing AOI must be located in the area of this extent, otherwise the area will be cut off by the extent boundaries. The processing area is calculated by the intersection of the image extent and the AOI. If the AOI doesn’t intersect the image boundaries the error will pop up.

Note

Image upload requirements (free plan):

• The file size must be less than 1 Gb

• Both sides image dimensions must not exceed 65535x65535 pixels

• The image must be georeferenced and the CRS must be one of: - WGS84 (EPSG: 4326) - Web mercator (EPSG: 3857) - UTM (any zone)

Hint

If your image doesn’t meet the params above, we suggest to use Mapflow API / QGIS plugin which have more capabilities. There you can also use our new My Imagery & Data Providers tool for storing and processing multiple images. Mapflow supports RGB imagery and also process single-band (panchromatic) imagery, but the AI models are not tuned for such kind of data, so the quality of the result may be worse than expected.

2.3 Choose from My Imagery

Instead of uploading a new file, you can reuse images you have already uploaded. Click Choose from My Imagery to open the picker and select a previously created mosaic or a single image as the source for the processing.

In the Pick from My Imagery dialog you can search your mosaics and images by name, sort them by creation date or size, preview the selected mosaic, and check its status. Select a mosaic and click Use this mosaic to set it as the imagery source. You can also create a new mosaic or upload images directly from this dialog.

Note

Read more about managing your uploaded imagery in My Imagery & Data Providers.

3. Select the imagery source

Note

If you uploaded GeoTIFF - the image source is already selected.

This tab allows you to select from:

• Imagery providers

• Custom URL

• Imagery Search

3.1 Imagery providers

Here you can select one of the default providers that are available in Mapflow, example:

• Mapbox Satellite is a provider of global high resolution satellite imagery. The date of the image and updates cannot be chosen.

• “Global mosaic” is a pilot version of mosaic of high res imagery (0.75–0.5 m/px) for year 2022. The preview is limited to zoom 15. Limited coverage for some countries. The mosaic is planned to be updated on a regular basis.

• ArcGIS World Imagery is a provider of the global coverage composed of high and medium resolution satellite imagery and aerial imagery, hosted by ESRI. The frequency of updating images is 1-5 years depending on the territory.

Hint

You can search the ArcGIS World Imagery metadata (date, zoom level) by location. To do this, use the Mapflow Imagery search tool in QGIS.

Important

Data providers are the TMS / XYZ data streaming services that can be connected to Mapflow to enable instant imagery analysis and AI mapping. Under the Mapflow commercial plans, we provide commercial providers access, historical imagery and updates on specific terms. We are continuously working on adding more commercial providers. For more details about the Imagery providers cost, see Mapflow subscription plans.

3.2 Custom URL

• Specify the XYZ link to the source of the images. The link is automatically checked for validity (for example, Openaerialmap - https://apps.kontur.io/raster-tiler/oam/mosaic/{z}/{x}/{y}.png);

• When adding you custom tile service it might be needed to define Y position: XYZ (top left), TMS/WMTS (bottom left);

• Provide your authorization data for the desired source (Login + Password);

• Add on map to preview.

3.3 Imagery Search

“Imagery search” allows Mapflow users to search for available satellite imagery over their area of analysis. It’s powered by Mapflow API providing access to the global satellite data providers through our partners.

See also

👉 See Imagery search for more information and guides.

4. AI model

Select one of the AI models (see Model description).

If your data source is Data provider (Imagery basemap), after selecting the model, the recommended zoom level for it will be set automatically. Note that you can always select any zoom level and check its availability with the “Suggest” button.

Warning

If the recommended zoom is unavailable, you can start the processing using lower zoom level, but the result may be less accurate.

4.1. Post-processing options

In this step, you can select additional processing options. Different AI models have their own options.

AI model “🏠 Buildings”:

• Classification - we currently recognize the following building types: apartment buildings, single-household dwellings, industrial, commercial, other non-residential (see Building Mapping classes).

• Merge with OSM - compare the results with the buildings in OpenStreetMap and, if a sufficient overlap has been found, use the ones from the OSM instead.

• Simplification - simplification of building contours.

• Regularization - regularization of building contours to make them more rectangular.

• Building heights - for each building, we estimate its height. If this option is selected, all roof contours will be shifted in accordance with their height, i.e. converted to “footprints”.

AI model “🌲 Forest”:

• Heights - classifies vegetation by height classes: by default 0-4 m for shrub (“low vegetation”), 4-10 m for regular forest (“medium forest”), 10+ m for area with tall trees (“high forest”). Classification is done per vegetated area, and not per single tree (see Forest Mapping classes).

• Tree crowns - extracts tree crowns from dense forest cover as well as free-standing trees.

5. Run the processing

At this step, you can set the name of your processing or use the initially generated one (the button allows you to generate variants of the name).

The “Start processing” will start processing.

The “Clear Selection” will clear the previously selected processing parameters.

Attention

Once you have selected the model and the processing parameters, you will see the total cost of your processing counted in Mapflow credits. Upon registration, you receive 250 credits for free for testing the platform (See Mapflow credits).

6. View the results

In the “Processing history” you see the progress of processing as well as view and get the processing results. Each running processing displays the selected model, the status, the creation date and the processing cost.

6.1 View on the map

After the successful processing completion, the card can be opened to view more detailed information about the processing parameters - AI Model, Post-processing, Area, Data Source.

Click the “View on the Map” to quickly view the processing result on the built-in interactive map.

Results Summary (Statistics)

Getting a vector layer is only part of the imagery analysis workflow. The next question is usually: what did the model actually find?

Every completed processing in Mapflow Web opens with an instant summary of its results – the number of detected features and their totals (area, length, counts) per output layer. The summary is calculated for the whole processing AOI, regardless of what is currently visible on the map.

Note

The service is powered by the Statistics API and is available for every default Mapflow model. Custom-model workflows include basic statistics, and their summaries can be customized as well (see Custom Models).

The summary panel

When the processing is finished, the panel appears above the map with the total number of detected features, the list of the output layers and the processed area:

The panel contains:

1. Header – the status of the dataset and the total number of detected features, e.g. “Dataset ready – 11,206 features detected”.

2. Subtitle – the output layers included in the processing and the size of the processed area, e.g. “crown · building · forest · roads · waterbodies over 1.00 km²”.

3. Per-layer statistics – one entry per layer, colored as the layer on the map, with the feature count and the corresponding total (area or length).

4. Zoom hint – tells you whether all the results are currently rendered on the map. At a low zoom level the map may display a part of the features, while the statistics are always counted for the whole result.

5. Actions – Download CSV and Ask about these results.

If you closed the panel, or opened an earlier processing, you can bring the summary back with the statistics button in the toolbar under the map.

What each model reports

Each model highlights the statistics that matter for its output:

MODEL	STATISTICS
🏠 Buildings	Number of buildings and their total area. With the Classification option applied, the counts are also given by building type (see Building Mapping classes).
🚗 Roads	Total length of the extracted road network.
🌲 Forest and trees	Total vegetation area. With the Tree crowns option applied, the number of the extracted crowns is reported as well.
🌍 Land use	Area per land use category.
💧 Water bodies	Number of the water bodies and their total area (the layer is added by the Water post-processing option).
[multi] models	One entry per output layer, e.g. crown, building, forest, roads, waterbodies.

Download the statistics as CSV

Click “Download CSV” in the summary panel to save the numbers as a table – one row per layer or class, with the counts and the totals. Use it to attach the figures to a report, or to compare several processings of the same area.

Note

PDF reports will be available soon.

Ask the Mapflow Agent about the results

Click “Ask about these results” to continue in the Mapflow Agent chat. The Agent instantly receives the structured statistics of the processing, so you can ask it to explain the numbers, compare the layers, or estimate the values you need, e.g.:

• “How much of the area is covered by vegetation?”

• “What is the average building footprint here?”

• “Summarize the results for a report.”

See also

View the results in the Get started guide, and the Mapflow Agent user guide.

6.2 Feature attributes and multilayers

Depending on the model and the options applied the extracted features might contain the semantic information that is written in the feature properties in GeoJSON.

E.g. “Buildings” model with “regularization” and “classification” by building typology options:

PROPERTY	DESCRIPTION
Area	Feature area in meters
Class_id	Buildings typology
Shape_type	The geometry primitive depending on the polygonization algorithm that’s been applied to the feature mask
Processing_date	The date of the processing completion
Simplification_score	The score is counted by the intersection of the pixel feature and polygonized feature. The higher score - the closer polygonized feature to the pixel one.

6.3 Layer Switching & Legend

You can switch the layers to preview on the map. Click the “sandwich” (☰) button on the top right corner of the map and extend the dialogue with the map legend.

E.g. “[multi] Buildings + Roads + Forest” model results with different layers:

6.5 Rate the processing

Important

We encourage you to rate the processing results. ⭐️⭐️⭐️ Your assessment is important for further improvement of the Mapflow AI models. We review all the user’s feedback and get back to you if any issues.

6.6 Share the results

You can share your project with the external users for the view-only access. Go to the Project – Share and enter the Mapflow user’s email.

Note

To upgrade project sharing rights and provide your collaborators with the editor’s access you need to switch to the Team accounts

7. Use the processing results

The Processing menu includes the items as follows:

• Click “Duplicate” on an existing processing to use its parameters as a starting point for a new one (on restart with the same parameters, new processing is started).

• Click “Source details” on the existing processing to view information about source details and AOI.

Tip

If you want to duplicate processing to change some params, like choosing another AI mapping model for the same AOI - use Duplicate

7.1 Download GeoJSON

Allows to download results as a file in GeoJSON format.

Warning

This option is not available in the free plan. Only paying users (see pricing) can use this option.

Working with API

Mapflow provides a REST API which, for example, allows you to query for currently running flows and fetch the results. If you are developing an application and want to use our API, - check out Mapflow Processing API.

Important

You must follow the requirements specified with Model requirements when uploading your own images for processing through the API of the Mapflow platform. Send a request using data preprocessing to help@geoalert.io.