Deploy Models
Deployment Options
This article applies to these versions of LandingLens:
After you are happy with the results of your trained model, you are ready to use it! To use a model, you deploy it, which means you put the model in a virtual location so that you can then upload images to it. When you upload images, the model runs inferences, which means that it detects what it was trained to look for.
| LandingLens | LandingLens on Snowflake |
|---|---|
Deployment Options
There are a few ways to deploy your LandingLens model:- Cloud Deployment: Deploy your model to a virtual environment hosted within the LandingLens app. Use API calls or to send images to your model. If using , the LandingLens app exists within your Snowflake environment, so the model will be hosted in Snowflake.
- LandingAI Deploy Docker: Download our Docker image to create a Dockerized container. Deploy your model and run inference in this self-hosted container. You must first activate a project to deploy its models with Docker.
- LandingEdge: Use the LandingEdge application to communicate with edge devices, industrial cameras, and programmable logic controllers (PLCs). You must first activate a project to deploy its models with LandingEdge.
| Attribute | Cloud Deployment | Container Deployment | LandingEdge |
|---|---|---|---|
| Latency | High | Low | Low |
| Throughput | Configurable | 30 FPS | 30 FPS |
| Pricing | Per inference | Free | Free |
For Enterprise users, can hide the LandingEdge and Docker deployment options if you don’t want your organization to deploy using these methods. To have these disabled, contact support@landing.ai.
When to Use Cloud Deployment
Cloud Deployment is a scalable and cost-effective deployment solution. It can accommodate surges in inference traffic up to a configurable rate limit, with charges incurred per inference. Cloud Deployment is a preferred option for managing variable inference loads. Use Cloud Deployment if you:- Want to start running inference without purchasing GPU machines or managing deployments.
- Have a good network connectivity from your inferencing point to the cloud.
When to Use Docker Deployment
Docker Deployment is the most flexible deployment option for developers that build mission-critical solutions or process high-throughput continuous inference loads. It can be deployed in your private cloud, on-premises, or at the edge. Use Docker Deployment if you:- Have a deployment infrastructure and want to add inferencing capabilities to it.
- Are looking for deployment automation in a container-based infrastructure.
When to Use LandingEdge
LandingEdge is an application that lets you deploy to an edge computer, such as an industrial PC. Use LandingEdge if you:- Want to build machine vision solutions using specialized hardware, like industrial cameras and PLCs.
- Want to build mission-critical solutions at the edge.
Compare Deployment Options
Use the table below as a reference when choosing a deployment option.| Feature | Cloud Deployment | Docker | LandingEdge |
|---|---|---|---|
| General | |||
| Hosting | LandingLens-hosted | Self-hosted | Self-hosted |
| Operating system | Linux, Mac, Windows | Any | Linux, Windows |
| Can run inference when not connected to the internet | ✖ | ✓ | ✓ |
| Can run inference on Visual Prompting models | ✓ | ✖ | ✖ |
| Can see live results in the user interface | ✓ | ✖ (there is no user interface) | ✓ |
| Can upload image metadata to images | ✓ | ✓ | ✓ |
| Maximum inference calls per minute | 40 | Depends on system | Depends on system |
| Can communicate with PLCs | ✖ | ✖ | ✓ |
| Can deploy on NVIDIA Jetson devices | ✖ | ✓ | ✖ |
| Can deploy on ARM64 processors | ✖ | ✓ | ✖ |
| Running inference consumes credits | ✓ | ✖ | ✖ |
| Send Images for Inference | |||
| Drag and drop images | ✓ | ✖ | ✓ |
| GenICam | ✖ | ✖ | ✓ |
| Images from webcam | ✓ | ✖ | ✖ |
| Video (will convert to images) | ✓ (via Python library) | ✖ | ✓ |
| Select from a designated folder (folder watcher) | ✖ | ✖ | ✓ |
| Send images via POST APIs | ✓ | ✓ | ✓ |
| Post-Inference Features | |||
| Apply post-processing scripts | ✓ (via Python library) | ✓ (via Python library) | ✓ |
| Can view inferenced images & predictions in LandingLens | ✓ | ✓ (must pass --upload flag) | ✓ (must enable “Upload results to LandingLens”) |
| Can save inferenced images to a local folder | ✖ | ✖ | ✓ |
The only deployment option for is Cloud Deployment.