top of page
Run Flai algorithems locally

Self-Hosted

Run Flai algorithems locally

Designed for scalability, Flai is adept at handling extensive datasets through efficient batch processing. Proven to process areas equivalent to an entire country or state, spanning tens of thousands of square kilometers, Flai solution can be deployed on cloud environments or local workstations. Flai's architecture is optimized for swift and efficient point cloud classification.

 

For sustained engagement, we work closely with clients to find the optimal business model that fits seamlessly into their operations. This comprehensive package includes dedicated support from the Flai team, ensuring optimal setup and maximum productivity.

Benefits of self-hosted solution 

 

  ✔ Classification speed

  ✔ Configurable batch processing

  ✔ Reduced Administrative Overhead

  ✔ Discounts and Incentives

  ✔ Security mitigation

  ✔ Priority Service

Server

Get trial license

Contact us at support@flai.ai to get started with a 10-day trial license. You have the option to utilize the Flai self-hosted solution without any restrictions.

Technical requirements

  • Operating system: Ubuntu 20.x or greater (LTS);

  • CPU: min 8 cores; optimally 16 cores;

  • RAM: min 32 GB; optimally 64 GB;

  • GPU: Nvidia Graphic card with minimum 11 GB vRAM; optimally 16 GB vRAM;

  • Disk: SSD disk with at least 500 GB of free space

​​​

​

Find more information about the requirements and installation process on Flai's Self-hosted documentation page.

​​

This comprehensive package includes dedicated support from the Flai team, ensuring optimal setup and maximum productivity. Custom AI models and retraining options are available, enhancing flexibility and ensuring a cost-effective, high-quality solution for point cloud classification. 

Testimonial​

​“Flai self-hosted tool seamlessly integrates into our workflow, providing accurate above-ground classification.
Its precise vegetation and building classification streamline our process, ensuring reliable and efficient data analysis.”

Marek Slowinski

CEO at

mobileallies.png
  • Is there a difference between the results processed on Flai web application and locally?
    The results are identical on the Flai web application and locally. If you process a flow online and then reuse it for local processing, you can expect the same results. The main difference is where the processing happens (online vs local) and that there is no need to upload/download the data when processing the data locally which results in faster turnaround time.
  • Do I need a computer with GPUs to run Flai self-hosted solution?
    Flai uses AI-based classification for LiDAR point cloud data, which is optimized for GPU processing.  While classification can also run on CPUs, it is up to 10 times slower than when processed on a GPU.
  • I want to process the same files multiple times to try out different settings. Will this be charged?
    No, processing of the same file only gets counted once. For example, if you process the same input file multiple times (different options, different model), it will only take it into account once. However, if you change the file in between then it is not considered as the exact same file and it gets counted next time you include it in the processing.
  • My coworker also wants to use the self-hosted version of Flai and run it on his projects. What should we do?
    If you have a valid license and you would like your coworker to use the same one, then it is best to add the person (as new user) to your company organization (link to docs). He should then generate a unique access token.
  • Can I run the self-hosted solution on Windows, or is it Linux-only?
    The Flai self-hosted solution requires a Linux (Ubuntu) operating system. However, it can also run inside WSL-2 (Windows Subsystem for Linux), meaning you don’t need a native Linux machine and can still use Flai solutions on Windows.
  • I want to know if Flai works as expected on the hardware setup I have
    There are many hardware configurations out there and it’s challenging for us to try all of them. However, we’ve prepared a benchmarking tool that allows you to run some preselected datasets and compare them with the stats we are getting (link)
  • How can I get the Flai self-hosted solution?
    To get a trial license, contact us at support@flai.ai. Once your license is activated, you can download the self-hosted executable from the Flai web app under the Flai CLI page. For installation, follow the step-by-step instructions on our documentation page.
  • What is the Flai self-hosted solution, and how can I use it?
    The Flai self-hosted solution is an executable that runs in terminal via a command-line interface (CLI). It uses the same processing algorithms as the Flai web app. The key advantage of the self-hosted solution is that it allows you to process LiDAR data locally eliminating the need to upload and process data in the cloud.
  • I cannot access the download of CLI tool or generate the access token. What's wrong?
    Reach out to support@flai.ai and request a trial license. Once the license is enabled by our team you will be able to download the CLI tool (link) and generate the access token.
  • What are the benefits of the custom AI service?
    With the custom AI service, you can train a custom AI model specific to your needs and extract specific objects from point clouds. This can help you to automate and streamline your workflows and improve the accuracy of your results.
  • What are the benefits of batch processing? What is the pricing like?
    For large scale projects batch processing might be beneficial. Flai can do the data processing on our infrastructure, in any cloud and / or on client’s infrastructure. The price for batch processing is based on the project size, point density, and the number of classified categories. You can contact Flai (support@flai.ai) to get a price estimate for your specific project.
  • How much I can process?
    You can process as much as you want and you will not be charged anything. We will only charge you when you would like to download the dataset. There are some system limits on how much you can process per month, probably you will not hit them, but if you will please contact us on support@flai.ai.
  • How much data can I upload?
    You can store up to 200 GBs of point cloud data in the application. If you would need to store more please contact us on support@flai.ai.
  • How is the pricing determined? Can you provide an example of how the pricing works?
    Pricing reflects the amount of processing units (PUs) that you consume, where 1 PU amounts to 0,01 EUR. Number of PUs consumed depends on 3 key factors: 1.) Point density The denser the point cloud (pts/m2) the higher the price 2.) Complexity of the use case (complexity of the flows) The more categories you classify and the more additional end products (i.e. vectorisation of objects in point cloud) you want to create the higher the price 3.) Area size The bigger the area (i.e. km2) the lower the price For more information, please contact us on support@flai.ai.
  • How is the price for batch processing calculated?
    The price for batch processing is based on the project size, point density, and the number of classified categories. You can contact Flai (support@flai.ai) to get a price estimate for your specific project.
  • How do you handle Security?
    When using our SaaS or PaaS services we prioritize data security at all times, ensuring that data in transit is always encrypted using HTTPS protocols. For data ingestion via FTP, we employ secure SFTP for an additional layer of protection. We utilize server-side encryption of data with Amazon S3-managed keys (SSE-S3) to secure data storage. This provides robust safeguards and keeps your data safe both in transit and at rest.
  • How can I estimate the cost before using the application?
    You can use the pricing calculator provided above to estimate specific project costs. However, to get precise pricing information it’s best to log into the web application, upload a dataset, create the desired flow and then the price will be displayed. For more information, please contact us on support@flai.ai
  • Is there a minimum charge or any additional fees associated with using the application?
    The minimum charge per download is 25 EUR. Other than that, there are no additional fees associated.
  • How often will I be billed for the usage of the application? Which payment options are accepted?
    You will be billed each time you download a dataset. Flai accepts credit card payments (online purchase) and payments via invoice. For more information, please contact us on support@flai.ai.
  • How does pay-per-download work? What is billable and what is not?
    For processing and viewing of the data you will not have to pay anything. You can explore and test the performance of AI completely free of charge. When you are ready to download the data then you will charged based on consumed processing units (PUs) of all dataset related flows.
  • I have multiple datasets (flow writers) inside the same flow, do I have to pay for each dataset separately?
    No. When you want to download a specific dataset you have to pay for the whole flow execution and therefore also all related datasets will be unlocked for downloading.
bottom of page