Software as a service (WebApp)

Point cloud classification

Perform semantic segmentation (classification) of Lidar point clouds using AI models. There are Advanced and Basic models available for each AI model. The categories you get in the end depend on your selection. You can choose pre-trained AI models derived from semantic segmentation datasets that are continuously labelled by a team of experts in the geospatial and forestry domain.

Tools for manual annotation

All manual edits can be used for active training of AI models. In other words, every manual correction that you do on the point cloud can be applied to future projects. Point cloud annotation can be done quickly with the use of intuitive tools. A cross-section view can be generated from each selection for a more accessible final selection. Annotations can be done in a perspective or orthographic 3D view.

Online 3D Viewer

Point clouds can be viewed directly in the web application. Adjust the display setting to get the best view:

• Perspective or orthographic projection,

• point size,

• point budget,

• display by attribute (elevation, intensity, classification).

Turning on or off the display of individual point classes is possible.

Processing flows

Each use case has its exceptions and requirements. Therefore we let you define the processing flow to best fit your project-specific needs. Workflow is expressed through an easy-to-use drag-and-drop interface. Different point cloud writers, readers and operators are available. Once the flow is determined, the processing is done automatically, and you only need to wait for results. No waking up in the middle of the night to start the next stage!

Object detection

The Flai web application tools can be used for object detection on Satellite data, images, and point clouds.

Flai's cloud machine learning platform can simultaneously be used for object detection on 2D data (images, satellite data) and 3D data (point clouds). For example, mobile laser scanning platforms simultaneously acquire images and Lidar, and with Flai web application, it is possible to use both data sources to extract objects. Satellite data can be used for the automatic detection of Flood areas and Dark vessels.

Automatically detected objects on images can be used for inventory, for example, railway infrastructure inventory, and car plates or faces detection and annomization. For example, point cloud feature extraction is used in autonomous driving applications.

AI learning point (Coming soon)

To train machine learning models, users have to provide training data. The generation of training data can be a challenging task. Flai's web application offers an AI learning point module that leverages active learning to cut down the needed time to annotate.

The system asks users to annotate just a few most informative examples and then retrains the AI models. The process can be repeated until satisfactory results are generated.

If you notice areas where the AI model does not perform optimally, use Tools for manual annotation directly on Flai's cloud-based machine learning platform.

Annotation-as-a-service is available if you would like us to do data labelling and annotation services to improve the quality and applicability of the AI model.