This post will look at an application of using ML and AI to analyze images of Strawberry plants for disease. I live in Florida, and Strawberries, not Oranges, is our #1 crop! Right now this is a very labor intensive process, workers must walk up and down rows of plants looking for diseased or damaged plants. This solution will show how the One Microsoft platform can be used to analyze plant images in real time and determine if a plant is diseased. This could help increase crop yields by greatly decreasing the time and labor needed to analyze the plants, and allow workers to focus on plants they already know have an issue. So the idea for teh solution is that drones or wheeled robots could be used to automatically capture images of plants in the fields, and then those images would be uploaded to Microsoft Onedrive for Business. A scheduled Microsoft Power Automate automated worker would then run and use an Image Classification model to determine which plants need attention, and could alert workers in real time with message like "Plant #76 in Row 234 appears to have mottle virus", and then the workers could direct their attention immediately to those plants that need human intervention. The solution uses the following Microsoft Platform components:
I published a series of videos on Youtube that details each step in the creation of the solution. if you're just interested in seeing the results, jump to step 4!
Before I came to work at Microsoft, I actually originally created this solution using a different automation engine, and Microsoft's Custom Vision tool:
https://www.customvision.ai/. The custom vision tool is very powerful, the only drawback to using it with Power Automate is you do have to do a bit of work to connect it, Lobe greatly simplifies this process. Here is the video detailing the Custom Vision Model:
Lobe,
https://www.lobe.ai/, is a recent Microsoft acquisition and has everything you need to bring your machine learning ideas to life. Just show it examples of what you want it to learn, and it automatically trains a custom machine learning model that can be shipped in your app. The big advantage of Lobe with Power Automate is it allows direct export to AI Builder! Here is a video that shows the creation and export of the Lobe image classification model:
Microsoft Power Automate,
https://docs.microsoft.com/en-us/power-automate/, empowers everyone to build automated processes with flows. Use low-code, drag-and-drop tools and hundreds of pre-built connectors that automate repetitive, mundane tasks with ease.
AI Builder,
https://docs.microsoft.com/en-us/ai-builder/overview, is a Microsoft Power Platform capability that provides AI models that are designed to optimize your business processes. AI Builder enables your business to use intelligence to automate processes and glean insights from your data in Power Apps and Power Automate. With AI Builder, you don't need coding or data science skills to access the power of AI. You can build custom models tailored to your needs, or choose a prebuilt model that is ready to use for many common business scenarios.
Here is the video detailing the building of the Power Automate automation:
Lastly, here is a link to the video showing the Automated Worker processing the uploaded images:
So what we have seen with this solution is how the One Microsoft platform allows the creation of powerful automations using AI that can help to decrease the amount of labor needed in crop disease detection, and thus hopefully increase crop output.
Comments
Post a Comment