In this quickstart, we'll use LandingLens — a Native App available in the Snowflake Marketplace — to create a computer vision model that detects pneumonia in X-ray images. After completing this quickstart, users can use the concepts and procedures from this quickstart to build Object Detection, Segmentation, and Classifications models in LandingLens.
LandingLens is a cloud-based Visual AI platform. LandingLens empowers users to create and train Visual AI models even if you don't have a background in AI, machine learning, or computer vision. LandingLens guides you through the process of uploading images, labeling them, training models, comparing model performance, and deploying models.
To users who are familiar with machine learning, LandingLens offers advanced tools to customize the model training process. LandingLens supports advanced deployment options including cloud deployment as well as Docker and LandingEdge, LandingAI's edge-deployment solution.
Access to the LandingLens app is available by request. To request the app, follow the instructions below:
After you've requested the app and been granted access it, follow the instructions below to install it in Snowflake:
Now that you've installed the LandingLens app, you are ready to get the sample images. LandingAI provides a set of sample images as an "app" that can be downloaded from the Snowflake Marketplace. You will use these images to train a computer vision model in LandingLens that detects pneumonia.
To get the sample images, follow these instructions:
Now that you've loaded the sample dataset into your Snowflake account, you're ready to create a computer vision model using those images in LandingLens.
Now that all of the images are in the LandingLens project and have classes assigned to them, train a computer vision model. When you train a model, you give the labeled images to a deep learning algorithm. This allows the algorithm to "learn" what to look for in images.
To train a model, click Train.
The right side panel opens and shows the model training progress. This process can take a few minutes.
Once training finishes, you will see the model's predictions and performance information. You can click the model tile in the side panel to see more detailed information. In most real-world use cases, you might need to upload and label more images to improve performance. In this example, the model should be performing well, so we will go to the next step, which is deploying the model.
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 inference, which means that it detects what it was trained to look for.
In this example, we're going to show how to use Cloud Deployment. You can also deploy models using LandingEdge and Docker.
To deploy the model with Cloud Deployment, follow these instructions:
After deploying a model with Cloud Deployment, a custom Python script displays at the bottom of the Deploy page. Copy this script, replace the placeholers with your information, and use the LandingLens Python library to integrate the model with your applications with very few lines of code.
Congratulations on creating a pneumonia detection computer vision model in LandingLens! You can now apply the concepts you've learned to building custom computer vision models in LandingLens.
In this quickstart you learned: