- 120
- 196 457
LandingAI
Приєднався 6 лют 2023
LandingLens on Snowflake: Overview and Demonstration
LandingLens on Snowflake is a native Snowflake application. LandingLens empowers users to create and train Visual AI models quickly. Watch this overview video to see how it works. The LandingLens Visual AI Platform guides you through the process of uploading images, labeling them, training models, comparing model performance, and deploying models.
Learn more:
landing.ai/snowflake
Get started:
app.snowflake.com/marketplace/listing/GZTYZ12K65D7/landingai-landinglens-visual-ai-platform
Video transcript:
Hello, I’m Andrea Kropp - a Machine Learning Engineer at LandingAI and it is my pleasure to give you a tour of LandingLens Native App on Snowflake. Everything you are about to see is within your Snowflake ecosystem.
LandingLens for Snowflake is the version of LandingLens that is available in the Snowflake Marketplace. So then what is LandingLens?
LandingLens is a Visual AI platform that streamlines the creation and deployment of computer vision models. With as few as 10 labeled images, users brand new to computer vision can train a visual AI model to classify images and detect and locate specific objects.
There are three stages in building a Visual AI solution - labeling, model training and deployment. In this demo you’ll see all three stages.
Here you are seeing the start of an object detection project to find bees in images.
Stage one is labeling the bees by placing bounding boxes around them. Having labeled 10 images, the Train button becomes available.
Stage two is training the model. In about a minute LandingLens trains a custom algorithm just for you with detailed metrics and performance reports.
Voila…the bee detector is done and ready to be put into service. On to Stage 3 - deployment.
There are three options for incorporating your visual AI algorithm into your business processes.
Option 1 - send an image to a cloud-based API endpoint. Here you see a new bee image being sent to the API endpoint via a one-line SQL command in a Snowflake worksheet.
Option 2 - use the LandingEdge local application to run inference without internet connectivity.
Option 3 - use the LandingAI Docker solution to integrate your computer vision model programmatically any place you can run Docker.
Using LandingLens natively in Snowflake has numerous benefits. You maintain the data privacy and governance of Snowflake, Are able to get fast access to GPU infrastructure via Snowpark Container Services, and can run Snowflake Copilot on your vision data to get instant insights.
We invite you to check out our fully functional free trial available in the Snowflake marketplace and start learning with the Landing Lens on Snowflake Fundamentals training course in our Community.
Learn more:
landing.ai/snowflake
Get started:
app.snowflake.com/marketplace/listing/GZTYZ12K65D7/landingai-landinglens-visual-ai-platform
Video transcript:
Hello, I’m Andrea Kropp - a Machine Learning Engineer at LandingAI and it is my pleasure to give you a tour of LandingLens Native App on Snowflake. Everything you are about to see is within your Snowflake ecosystem.
LandingLens for Snowflake is the version of LandingLens that is available in the Snowflake Marketplace. So then what is LandingLens?
LandingLens is a Visual AI platform that streamlines the creation and deployment of computer vision models. With as few as 10 labeled images, users brand new to computer vision can train a visual AI model to classify images and detect and locate specific objects.
There are three stages in building a Visual AI solution - labeling, model training and deployment. In this demo you’ll see all three stages.
Here you are seeing the start of an object detection project to find bees in images.
Stage one is labeling the bees by placing bounding boxes around them. Having labeled 10 images, the Train button becomes available.
Stage two is training the model. In about a minute LandingLens trains a custom algorithm just for you with detailed metrics and performance reports.
Voila…the bee detector is done and ready to be put into service. On to Stage 3 - deployment.
There are three options for incorporating your visual AI algorithm into your business processes.
Option 1 - send an image to a cloud-based API endpoint. Here you see a new bee image being sent to the API endpoint via a one-line SQL command in a Snowflake worksheet.
Option 2 - use the LandingEdge local application to run inference without internet connectivity.
Option 3 - use the LandingAI Docker solution to integrate your computer vision model programmatically any place you can run Docker.
Using LandingLens natively in Snowflake has numerous benefits. You maintain the data privacy and governance of Snowflake, Are able to get fast access to GPU infrastructure via Snowpark Container Services, and can run Snowflake Copilot on your vision data to get instant insights.
We invite you to check out our fully functional free trial available in the Snowflake marketplace and start learning with the Landing Lens on Snowflake Fundamentals training course in our Community.
Переглядів: 45
Відео
VisionAgent in Action
Переглядів 695Місяць тому
Dr. Andrew Ng and the LandingAI Machine Learning Engineers demo VisionAgent, the agentic workflow solution from LandingAI. Check out these VisionAgent resources: - VisionAgent web-based tool: va.landing.ai/ - VisionAgent repo: github.com/landing-ai/vision-agent - Discord: discord.com/channels/1011796394090168390/1011796394090168393
Hands-On Lab: Mastering Visual AI
Переглядів 379Місяць тому
Watch this immersive, hands-on webinar where you'll explore LandingLens's practical, cutting-edge features. This session is designed for professionals seeking to leverage the latest advancements in visual AI technology to enhance their workflows and drive impactful results. In this interactive lab, you'll explore LandingLens's newest capabilities, including advanced model training, enhanced dat...
Improving EV Battery Inspection Using Advanced Imaging & AI
Переглядів 3433 місяці тому
The growing market for electric vehicles is driving demand for different battery technologies. While each battery is designed to meet the specific needs of an application, all batteries must be lightweight and compact, have a long life both in use and in storage, and deliver a relatively consistent voltage during operation. While these goals can be achieved in numerous ways using different batt...
Vision Agent: Agentic Program Generation for Vision Tasks
Переглядів 8923 місяці тому
In this video we show our latest version of Vision Agent that can generate python programs to solve vision tasks. This allows the you to re-run the program many times again without ever having to make expensive and slow calls to LLMs or vision-LLMs to do repetitive tasks. We also show an example use case where Vision Agent generates a program to locate when surfers are close to sharks. Join Ou...
Snowflake and LandingAI Collaborate to Make Complex Diagnoses Easy
Переглядів 7604 місяці тому
When great partners come together, magic happens! ⚡ ️ Snowflake and LandingAI are collaborating to make complex medical diagnoses easy. In this video, Jeff Hollan, Director of Product at Snowflake, shows us how to go from images to interactive data (powered by natural language) in a few simple steps. Through the Snowflake and Landing AI partnership, Snowflake users can leverage the computer vis...
Leveraging Large Vision Models for Life Sciences (from R&D through Commercialization)
Переглядів 2534 місяці тому
Step into the world of medical devices and pharmaceutical technology, where upholding safety, efficacy, and quality standards is non-negotiable. Discover how AI-driven computer vision revolutionizes inspection and defect detection in manufacturing, ensuring unparalleled reliability. Delve into the transformative potential of AI for the medical devices and pharmaceutical industries. Join Erik Tr...
Transform video data into JSON with Vision Agent
Переглядів 6674 місяці тому
In this video, We demonstrate how to use a vision agent to analyze a video and identify if a child is dangerously close to a pool. We explain the process of extracting frames, running a grounding model, using OCR to get timestamps, and calculating the distance between the child and the pool. The goal is to transform a video which is unstructured data to a JSON which is structured data. We also ...
Computer Vision Workshop: MLOps Best Practice
Переглядів 4694 місяці тому
View this webinar hosted by Erica Gilbert, Custom Success Engineer at Landing AI. Erica walks you through how to ensure your computer vision model is ready for deployment. She shows you how to use key features from LandingLens to power your process! You'll learn how to use label agreement to determine SME accuracy, perform custom model training, and use model comparison tools for consistency in...
Download Models and More! Pricing & Feature Updates, May 2024
Переглядів 2174 місяці тому
Learn about the new features in LandingLens, the computer vision platform from Landing AI. Updates include: - All users can use LandingEdge and Docker deployment, no license required! - All users can DOWNLOAD MODELS (yes, this means you can access the ONNX file) - The number of credits used by Custom Training is now tied to the actual training settings Want to share your use case or have questi...
Run Vision Agent with your tools
Переглядів 7424 місяці тому
In this video, we demonstrate how to use vision-agent to find the most prominent item in a visual scene and then count it, specifically in a retail setting. We show how we created a custom tool called “Store Item Deduction” and explain its usage. We also discuss the planning, execution and explanation of certain parameters in vision agent. We run the end to end demo on two images and discuss th...
Processing Baby Cam Recording with Vision Agent
Переглядів 3065 місяців тому
In this video, I’ll show you how I leveraged the Vision Agent library to find out where my daughter sleeps-whether it's in her crib or on the main bed, and walk you through what happened in a vision agent execution. Join Our Community 🤖 Interested in vision agent technology? Join our Discord community to share ideas, get support, and collaborate on the vision agent library. Whether you're a se...
Improving Semiconductor Defect Detection & Classification Using Large Vision Models (LVMs)
Переглядів 1,2 тис.5 місяців тому
Semiconductor and MEMS manufacturers around the world are maximizing their yield and reducing expenses through the use of AI. These manufacturers empower advanced technologies and elevate everyday experiences. Watch this recorded webinar to learn how Domain-Specific Large Vision Models (LVMs) can provide a step function increase in the time to value and wafer inspection accuracy. Hear from Land...
Accelerate AI Deployment with Domain-Specific Large Models (LVMs) in Energy & Utilities
Переглядів 3135 місяців тому
Computer vision is widely used in the Energy and Utilities sector such as Asset Inspection, Condition Monitoring, Predictive Maintenance, Load Monitoring & Optimization, Infrastructure Planning, and Vegetation Management. Discover how you can accelerate AI deployment in Energy and Utilities with Domain-Specific Large Vision Model with Marco Fernandez, Director of Engineering at Capgemini, and C...
Developing and Deploying Computer Vision Applications
Переглядів 4185 місяців тому
Join our Technical Deep-Dive Office Hours to acquire the skills to develop and deploy custom computer vision applications effectively for specific projects and different enterprise and industry use cases.
The Landing AI Machine Learning Engineering Team Discusses “Agent Frameworks”
Переглядів 3636 місяців тому
The Landing AI Machine Learning Engineering Team Discusses “Agent Frameworks”
Manage Data Transforms: Rescale, Resize, & Crop
Переглядів 1036 місяців тому
Manage Data Transforms: Rescale, Resize, & Crop
Run Parallel Training in Custom Training
Переглядів 1067 місяців тому
Run Parallel Training in Custom Training
Landing AI Office Hours: Feb. 21, 2024
Переглядів 1507 місяців тому
Landing AI Office Hours: Feb. 21, 2024
Technical Deep Dive - Integrating Computer Vision into your Applications
Переглядів 5067 місяців тому
Technical Deep Dive - Integrating Computer Vision into your Applications
The Future of Inspection: How AI and Large Vision Models Advance Industry Inspections
Переглядів 9727 місяців тому
The Future of Inspection: How AI and Large Vision Models Advance Industry Inspections
Landing AI Office Hours: Jan. 24, 2024
Переглядів 1488 місяців тому
Landing AI Office Hours: Jan. 24, 2024
Refer to Heat Maps for Classification Models
Переглядів 2998 місяців тому
Refer to Heat Maps for Classification Models
Domain Specific LVMs Training Session
Переглядів 1,9 тис.8 місяців тому
Domain Specific LVMs Training Session
nice cat
TY 4 Vid
can we do web browsing using vision agent
Cool ! Will gonna try :)
why generate a program? should this not work like calling gpt in a loop like other agent frameworks?
like autoGPT? I think they're calling GPT in a loop based on the number of GET requests to openai. But they're saying outputting python is cheaper because you don't need to call openai api when you run the program on another video later
I Participated in the seminar, I must say, i am really impressed by the way the seminar was structured. I had a lot of questions that were answered effectively. Thanks Adrian for inviting me.
Nice thank you
LOL did you notice the file he had in his folder? ☠
Ng i love you. use a teleprompter for best video presenter to audience connection.
Love this guy Ng
Hi, I worked on OCR project back in 2020, and the most common tool used back than, as per my knowledge, was Tesseract OCR. It was very brittle with regard to language and fonts and a lot of heuristic preprocessing was need. But my use case was a few shot, no training was applied. I just wonder where the field is today, are the tools much better now?
Hi, @ninatko. The field has advanced a lot with deep learning based approaches. I have also used Tesseract in the past, and it's true that these classic approaches seem to be very brittle. However, I know that Tesseract added a LSTM deep learning base model since version 4. Some advantages of the newer deep learning approaches, like the one we offer at Landing AI is the ability to locate the position of the different texts in the image, as well as making them more robust under different conditions. However, sometimes you will still need some heuristic preprocessing, depending on your data. Hope that helps.
@@CamiloLandingAI Thank you for the reply! Interesting that they use LSTM, would not a transformer do a better job, are LSTMs better for characters in contrast to subwords/tokens? OCR is based on characters and not tokens as the modern language models are, right?
Thank you for raising this topic. It is super insightful. As I am currently doing research regarding AI for inspection, is it possible to have further discussion with Landing AI team? Thank you
Very interested in this... can I make a suggestion. Turn off monetization to eliminate advertisements while watching.
very informative
Could you share the source of the original work on the domain specific LVM illustrated at 1:31?
Hi, do you plan on showing how to do labelization from an external front end ? I'm wondering how I could implement this functionnality from another frontend, and send back the data to landingAI
Excellent marketing :-D What is the advantage over a very specific model that has been trained with 10-30 examples? The accuracy is much higher, the effort is much lower (about 20 min) and the model is much easier to control and change. Also, as you correctly said, industry examples are very specific and the cases are often very defined.
I am a teacher help me to teach my student
🎉 Saludos desde Valledupar Colombia.
Give please the examples of large vision models ?
What's that semi conductor thing
Great info, thanks alot!!!
This person is an idol to me ❤
Same. However is video feels like its generated :D Is it @LandingAI?
@@lukeskywalker7029 yes
Man. I was thinking about starting modular aquaponic systems for urban households. Would like to provide fishsafety as a service. And maybe also connect it to thermalimages. Very informative hopefully they will be available in the cloud soon. X) 🐙🙂😅🆒🆗😁
Insightful💡
Awesome presentation, Thanks, Dr. Ng.
😍😍😍
Nicely explained !!
Nice job Andrew! I've worked with Eddy Shyu as an alpha tester of your prompt engineering course, and have been going through several of your courses for almost a year now. Looks like you innovated again! Great work!
Hey Team, Yesterday there was a session with Andrew NG on your platform. I missed it. As a beginner CV developer, I eagerly waited for that. But unfortunately I can't attend bcoz of the client meeting. Can you guys, please share the recordings on utube please. Thanks in advance, Team.
awesome!!
Our future can't exist without ai.
Is there a reference paper for this tool or a similar one? Thank you.
Great stuff
In case I want to count the bacteria colonies in petri disc. Can it work using landing ai?
Hi @gusmein5144 - You can convert the Visual Prompting masks to COCO format to get the bounding box count. However, this might get tricky if the bacteria colonies are side by side or overlapping. We'd be happy to help you with your use case. If you'd like to chat with us or if you have any additional questions, feel free to create a post in our "Ask the Community" forum in the Landing AI community: community.landing.ai/c/ask-the-community/.
Very useful idea Sir. Is it possible to "after labeling and then counting the amount of colonies in the image using visual prompt? Please make a video counting object after labeling images using visual prompt. Thanks Sir.
Can you make a sample video for python?
nice tool
Super - Looking forward to some exploration of LandingLens.
I try this project but it didn't work.
Hi! I'm sorry to hear the project wasn't working for you. Instead of cloning the repo, try referencing the collab notebook here: colab.research.google.com/github/landing-ai/landingai-python/blob/main/examples/webcam-collab-notebook/webcam-collab-notebook.ipynb#scrollTo=98KX-kH9ZdKs
Great news!! 🎉
تكنولوجيا حديثة جدا 👍💕💕💕💕
This thing looks interesting
'PromoSM' 🎶
I want to know why you never uses high quality videos even in his lecture courses
Thanks for this
You're welcome, Zachary!
Thanks for the concise demo.