Are you curious about how to turn a live camera feed into actionable insights in real time without writing a ton of code or setting up a complex infrastructure?
In this hands-on demo, Andrea Kropp, Machine Learning Engineer at LandingAI, walks through the process of using Snowflake and LandingLens, a Snowflake Native vision AI app, to connect to a YouTube livestream, detect vehicles, such as passenger cars, trucks, and school buses, and visualize the data in a real-time dashboard built with Streamlit. No GPU setup required. She also shows how this technology can be used for such real-world applications as monitoring doggy daycares, city signage in Thailand, and even wildlife in the Namib Desert.
You’ll learn how to:
- Collect and label training images from a video stream
- Train a custom object-detection model using LandingLens
- Deploy the model to an API endpoint
- Send images from a live feed for inference every 10 seconds
- Store structured results in Snowflake tables
- Build interactive analytics and visualizations in Streamlit
- Expand this setup to other use cases, from traffic monitoring to wildlife detection
This video is ideal for:
- Data scientists and ML engineers
- Snowflake users exploring native apps
- Developers and solution architects
- City planners, transportation analysts, and anyone interested in video analytics
- Anyone curious about bringing computer vision into a no-code/low-code workflow
Learn more about LandingLens and try out some sample code using the same livestream feed by visiting https://landing.ai/snowflake
Try out LandingAI’s LandingLens Native app in Snowflake’s Marketplace.
