Flypix

What is FlyPix?

FlyPix is a code-free geospatial AI analysis platform developed by FlyPix AI GmbH, with its headquarters located in Darmstadt, Germany. It is designed for satellite, aerial, and drone imagery, bringing data labeling, model training, object detection, measurement, and map visualization together in the same workspace.

The platform is designed to address the issue of low efficiency associated with manually checking large numbers of images one by one. Users do not need to write any model code in order to create specialized detection models for buildings, vehicles, vegetation, water bodies, infrastructure, or other objects.

Main functions

Code-free custom models

Users mark a small number of target samples on the images to define the categories to be recognized, after which the platform trains a custom model. This approach is suitable for agricultural, engineering, mining, and environmental tasks where the specific targets are not available in general model databases.

Object detection and localization

The model can detect, segment, classify, and locate objects in dense areas or large expanses of terrain. The results include attributes such as geographic location, category, area, size, and quantity.

Intelligent annotation

The annotation tool supports various forms of training data such as box selections, polygons, and segmentation masks, and it enables reduced manual effort through automated annotation. However, humans are still required to review any missed detections, incorrect detections, and issues related to the quality of the boundaries.

Change and anomaly detection

After users upload images of the same area taken on different dates, it is possible to identify objects that have been added, disappeared, moved, or whose scope has changed. This approach is suitable for monitoring construction progress, forest changes, disaster assessment, land use, and facilities.

Measurement and Statistics

The platform can count the detection results and calculate areas, lengths, distances, and other spatial metrics. The accuracy of the measurements depends on the image resolution, geographic registration, and the quality of the model boundaries.

Maps, heatmaps, and dashboards

The analysis results can be viewed on interactive maps, heat maps, and dashboards, allowing for the comparison of regional patterns, trends, and object density. Teams can transform technical detections into more accessible business-oriented representations.

Multi-source remote sensing data

FlyPix supports data types such as RGB, multispectral, hyperspectral, LiDAR, and synthetic aperture radar. Different sensors can be used for capturing information on color, materials, vegetation, terrain, as well as for conducting observations in all weather conditions.

Image data market

After logging in, you can browse, preview, and request high-resolution images from various providers in the integrated data market. The costs associated with purchasing such images, as well as the licensing terms and coverage areas, need to be verified separately from the platform subscription fees.

Collaboration and GIS export

The paid tiers allow for sharing maps, inviting collaborators, and importing or exporting spatial data such as GeoJSON. Advanced plans also offer options for access control, APIs, white labeling, and local deployment.

Supported data

The specific file extensions allowed, the maximum size of a single file, the coordinate reference system, and the requirements regarding bands should be determined according to the importer tool within the application. Just because a certain data type is supported does not mean that any file that lacks registration or metadata can produce reliable results.

Complete workflow

1. Register an account and select a plan that matches your data volume and collaboration needs.
2. Create a project, upload your own images, or obtain images from data markets.
3. Check the coordinates, resolution, date of capture, cloud cover, and data usage license.
4. Define the categories of objects to be detected, and designate a set of representative positive and negative samples.
5. Train a custom model, and view the validation metrics and sample predictions.
6. Run detection, segmentation, measurement, or change analysis in the target area.
7. Manual review of false positives, missed detections, as well as the geometry of adjacent objects and boundaries.
8. Aggregate quantities, areas, and time changes on maps or dashboards.
9. Export vector layers or integrate with existing GIS processes via advanced interfaces.
10. Save the model version, data date, and thresholds to facilitate subsequent reruns and audits.

Methods to improve the quality of analysis

• The training samples cover different locations, seasons, lighting conditions, and sensors.
• Also mark backgrounds that are easy to confuse and similar objects.
• Large objects are enclosed by complete borders, while small objects are designed so that too much background is not included within their frame.
• Split the training and validation data by region to reduce spatial leakage.
• Use manual sampling to calculate accuracy, recall, and margin error.
• For important decisions, keep the original images rather than only saving the AI-generated results.

Price packages

The current pricing system has been switched to a dynamic approach; some older pages on the official website still display the options Basic and Professional along with old euro amounts, which do not match the points and storage options offered under the new system. Do not consider the prices shown on those old pages as the current prices – instead, refer to the settlement page after logging in or to the sales quotes provided.

Annual discounts, taxes, additional seats, video purchases, and extra storage may be charged separately. Before signing the contract, it is necessary to determine how points, model training, the data market, and API calls will be billed.

How to calculate the integral

The current pricing is based on the amount of pixels processed, with 1 credit equivalent to 20 MP. The page indicates that 1 credit equals 1 euro, but the actual cost per credit in the context of a subscription needs to be determined by considering the price of the respective plan.

• For the same area of ground, the higher the resolution, the greater the number of pixels and the more energy is typically consumed.
• When processing centimeter-level images from drones, more pixels are required for the same area compared to low-resolution satellite images.
• Training, inference, and some advanced functions may consume points.
• If the points are insufficient, top-ups are available; however, the price and validity period of such top-ups depend on the account settings.
• It should be confirmed before settlement whether monthly points can be carried over and whether failed tasks will be refunded.

Suitable for users and scenarios

• Agricultural team: crop counting, health monitoring, irrigation, and disease detection.
• Construction and infrastructure: site progress, asset identification, and inspections.
• Urban planning: analysis of land use, architectural changes, and regional development.
• Environmental agencies: Monitoring of forests, water bodies, habitats, and ecological changes.
• Energy and Mining: Facility inspection, site changes, and risk management.
• Ports and logistics: Identification of ships, containers, and site objects.
• Disaster response: Assistance in identifying the affected areas, the assets that have been damaged, and the allocation of resources.

Advantages

• Bring annotation, training, inference, visualization, and export together in one platform.
• The code-free interface lowers the barriers to training specialized models for GIS teams.
• It covers optics, multispectral, hyperspectral, SAR, and LiDAR.
• It supports the expansion from small-scale drone projects to large-scale satellite analysis.
• Advanced plans offer APIs, white labeling, on-premises deployment, and professional support.

Restrictions and risks

• The test results are not absolutely accurate; clouds, shadows, season, and resolution can affect the outcomes.
• A small number of annotations are sufficient to initiate training, but complex scenarios generally require a larger volume of samples.
• The results regarding area and distance depend on accurate geographic registration and pixel scale.
• Comparisons of website speeds are just specific examples and cannot replace one’s own benchmark tests.
• Dynamic pricing is in conflict with the versions of the old marketing pages; written confirmation is required for purchases.
• Advanced APIs, on-premises deployment, and SLAs require customized solutions.
• Critical security, law enforcement, or disaster-related decisions must be reviewed by professionals.

API, integration, and open-source status

The fact that FlyPix participates in partner programs or uses open-source components does not mean that the platform itself is open source. When preparing for API integration, it is necessary to obtain information regarding authentication methods, object models, asynchronous tasks, rate limiting, error codes, and data retention policies.

Privacy and security

FlyPix ensures that data is processed in accordance with the principles of GDPR, and protective measures are taken for both data in transit and static data. The privacy policy applies to both websites and applications; users can request access to or deletion of their personal data.

The platform records the IP address, browser type, internet service provider, time, URL visited, and click information, and it uses cookies for Google login. The terms also permit the use of anonymized, aggregated, or de-identified data to improve the system.

• Before uploading, verify the permissions for the images, personal information, and sensitive facilities.
• The government and key infrastructure projects should verify where the data is stored and with whom subcontractors are engaged.
• The contract specifies the ownership of the project data, models, and export results.
• It is required to define the procedures for backing up data, deleting it, notifying in case of incidents, and closing accounts.
• For on-premises deployment, it is still necessary to evaluate model updates, telemetry, and the scope of support access.

Terms, Copyrights, and Refunds

Users must have the legitimate right to upload data, and must not commit any acts of infringement or violation of the law, nor must they exploit the service or bypass security measures. The service is provided as it is, and the platform does not guarantee that the results will always be accurate, reliable, or uninterrupted.

The terms allow for adjustments to prices or features, as well as the discontinuation of certain services, and they limit liability regarding third-party tools. The terms mention a return policy, but there is no clear commitment regarding refunds for subscriptions; the conditions for cancellation and refund must be confirmed in writing before making a purchase.

Frequently Asked Questions

Is FlyPix a regular image recognition tool?

No. It is designed to handle remote sensing and surveying data that contains geographic information, and the results obtained can be used for creating maps, conducting measurements, analyzing changes, and generating GIS layers.

Is it necessary to know how to program or train neural networks?

It is not necessary. Users upload data and label samples through the interface and initiate training, but it remains important to understand remote sensing data and quality assessment.

Which sensors can be processed?

The platform officially supports RGB, multispectral, hyperspectral, LiDAR, and SAR data, as well as satellite, aerial, and drone data.

What are the limits in the free version?

It is now possible to create an account without a credit card, but the free version’s features such as real-time storage, points, and other benefits require an account to be established. The 3GB storage space and 10 points mentioned on the old page should not be considered as guarantees under current terms.

How much data can one integral handle?

The current conversion is approximately 20 MP per 1 point. The area that can be covered depends on the resolution of the image; therefore, high-resolution drone data requires more points.

Can it be exported directly to GIS?

The paid plan supports import/export in formats such as GeoJSON, as well as workflows for vector layers. Specific attributes, coordinate systems, and other formats need to be verified within the account.

Are APIs provided?

It is available, but according to the current pricing structure, the API is included in the Advanced package. There is insufficient information available regarding the public SDKs, endpoints, and rate limits; therefore, it is necessary to contact sales for the technical documentation.

Can it be deployed on one’s own server?

For Advanced and Enterprise solutions, local deployment can be discussed; hardware, upgrades, security aspects, and costs require a customized assessment.

Can the test results be used directly for decision-making?

It is not recommended to use the model directly without further verification. An independent validation set, manual sampling, and confirmation by industry experts are needed to assess the errors of the model in the target area and on the sensors.

Summary

FlyPix integrates multi-source remote sensing data, code-free model training, object detection, change detection, spatial measurement, and GIS export into a cohesive workflow. It is suitable for professional teams that wish to reduce reliance on manual interpretation without having to set up their own machine learning infrastructure.

The selection criteria involve not only functionality but also pixel integration, data acquisition, geographical accuracy, privacy considerations, and model validation. Due to discrepancies between the old and new price lists, the official purchase price should be based on the dynamic pricing page or the written quote from FlyPix AI GmbH.