Academy
In this tutorial video, discover four ways of evaluating the performance of a Classification computer vision model in LandingLens on Snowflake: Model metrics such as F1, Precision and Recall Confusion matrix, confidence scores and visual inspection of misclassifications Heatmap overlays on image pixels Evaluations sets and side-by-side comparisons
In this video tutorial, you will see how to configure four computer vision models and begin model training jobs in LandingLens on Snowflake. Learn about data partitioning, GPU provisioning, and model training configurations. Monitor training progress through loss charts.
Bring images into a Snowflake stage and sync the contents with LandingLens on Snowflake. This tutorial uses pre-labeled images for a binary classification project. It shows how to create the stage in Snowsight and how to grant the required permissions for the image sync to succeed.
When your model encounters something new, it may be time to retrain. This video shows how a model in production is confused by an object it has not seen before – horses. In LandingLens it is easy to create a new class, label a small number of images and retrain the model to detect […]
Assign images to be labeled to one or more labelers. Create a labeling task for a set of images and choose how many labelers should complete each task. Track the team’s progress labeling the assigned images for your machine learning or computer vision project.
LandingLens has built-in tools to assist with labeling images that have poor contrast. Auto-contrast and histogram equalization are especially helpful for low-light and infrared images used in machine learning and computer vision.
See how to apply the concept of continuous learning when working with LandingLens and LandingEdge. Mark images that have been sent for inference correct or incorrect with human judgement. Then use the incorrect images to improve your model. You can automatically add images that you’ve run inference on using LandingEdge into your dataset in LandingLens, […]
LandingLens’ Custom Training process lets you customize certain training settings, including data transforms. Transforms let you rescale, resize, and crop all the images in your dataset before model training starts. Take a look at this video to learn more.
LandingLens offers parallel training for Custom Training! Parallel training is the process of training multiple models simultaneously. This is helpful if you want to compare performance between model training settings. The Custom Training process is designed for users who are familiar with machine learning concepts and understand how the different settings impact the resulting model. […]
Get LandingAI's Monthly Newsletter
Stay updated with the latest computer vision and AI news and resources delivered to your inbox.