Train Models
  • 24 May 2023
  • 4 Minutes to read
  • Dark
  • PDF

Train Models

  • Dark
  • PDF

Article Summary

LandingLens offers two Model Training options:

  • Fast Training: This is the default option. LandingLens will automatically fine-tune its settings to optimize Model Training speed.
  • Custom Training: Also called "Advanced Training". This is an advanced option recommended for users who are familiar with machine learning. Custom Training allows these users to have more granular control over pre-processing transforms and manual hyperparameter tuning. For more information, go to Custom Training.
The information in this section is not applicable to Visual Prompting. For more information, go to Visual Prompting.

Images Are Resized During Model Training

When you train your Model, LandingLens resizes all images to make training faster and more effective. This resize does not affect the original images. This means that the original high-resolution images are always available in the Data Browser.

When you use Fast Training (the default Train option), all images are auto-resized according to your Project Type.

  • Classification: Resized to 512x512px 
  • Object Detection: Rescaled with padding to 640x640px 
  • Segmentation: Resized to 800x800px
  • Anomaly Detection: Resized to 512x512px
If you have large images that include small objects to detect, these objects can be lost due to the automatic resize. You can use Custom Training to choose the resize width and height manually. Or, you can resize images before you upload them to your LandingLens Project.

Train and Save Models

  1. Open the Project you want to train.
  2. Click Train. This starts the Fast Training process. (If you're familiar with machine learning and would like to configure advanced settings, go to Custom Training.)
    Use the "Train" Button for Fast Train
  3. LandingLens runs the Model Training process and creates a Model.
  4. If you want to keep the Model and refer to it again, click Save. If you don't save a Model, it is overwritten the next time you perform training.
    Save the Model
  5. Provide a short, descriptive name for your Model in the Model Name field. For example, you can enter the Precision and Recall results.
  6. Click Save.
    Name the Model
  7. You will be prompted to create an endpoint and deploy the Model. If you're not ready to deploy, click the to close the pop-up window.
    Close the Window

Model Training Process

Models go through a series of steps when they train. This section describes those steps at a high level. 

  1. Provisioning GPU. LandingLens will warm up a virtual computer in the cloud that includes a graphics processing unit (GPU). This virtual computer is where the Model Training occurs.
    Step 1: Provisioning the GPU
  2. Configuring Your Model. The platform will send the Model's data and details to the virtual computer. This information will tell the Model how to train itself.
    Step 2: Configuring Your Model
  3. Training & Learning. An error curve will display. This graph will show how well the Model is training. If the line drops when it trains, the Model is making fewer errors and is learning to be more accurate. The Model will test its accuracy by comparing its Predictions to your Ground Truth.
    Step 3: Training & Learning
  4. Calculating Model Performance. The platform will load the metrics and show you how well the Model performed. Go here for more information on these metrics.
    Step 4: Model Performance

View Trained Models Reports

After you train a Model, LandingLens creates a report for that Model. To view these reports:

  1. Open the Project.
  2. Click Models at the top of the page.
    Open Models
  3. Click the ellipses icon (:) next to the model name.
    See More Information About Models
  4. View the displayed report.

Model Report Details

The report is divided into different sections, as described in the table below.

Trained Model Report 

1Training Info
  • Save a Model by clicking the pencil icon and entering a title for the Model.
  • View the status of the Model, who trained it, and when it was trained.
2Learning CurveData is sent to your Model when it trains. While the Model trains, LandingLens displays a graph that shows how well the Model is training. This graph is similar, except that it also includes a timeline that shows how long the Model took to train.

When you train a Model, a virtual computer in the cloud warms up. Because of this step, the starting time does not start at 0 seconds. Instead, the graph starts after the amount of time it took to acquire the virtual computer.
Go here to learn more about the Model Training process.
3ConfigurationThis information provides more details on your Model training. For example, you can see how many Epoch cycles your Model went through.
4LogsThis information is helpful for troubleshooting. If you experienced any issues during Model Training, you can provide these logs to a Landing AI associate. 


What happens when I stop Model Training? What are the results?

When you stop training, LandingLens completes its current round (Epoch) of training and evaluates the images in your dataset with the Model it had generated up to that point. The results displayed are of the "best" Model at that point. 

How does Fast Training work?

LandingLens fine-tunes its settings to optimize Model Training speed.

Was this article helpful?