result of a plant analysis with good result
welcome screen prompting to select the plant image
result of image analysis with a bad result
chatbot providing recommendations for a better plant care
unhealthy leaf because of nitrogen lack
Gardening is hard. Everybody who has plants at home knows that very probably they are going to die in the next few months. So, we created PlantBot. The first automated bot that diagnoses the status of your plants. It is really easy to use plant bot on your plants. You just have to upload a picture from your phone and we will analyze it using Watson. If based on that picture we detect problems on your plant our chat bot will ask you a set of questions to find out what are you doing wrong. No more plants without the right cares.
How I built it
We created a mobile application that enables to take picture to your plant. When you take a picture be process it with our knowledge base on Watson. We have trained this knowledge base to detect actual problems of plants based on more than one hundred pictures of different healthy and unhealthy plants. Moreover, based on the crowd sourced feedback and text analysis we can provide the right feedback to know what is the actual problem in your plants.
Challenges I ran into
The main challenge has been creating the right knowledge base so Watson can do the right analysis on images and working with the conversational API.
Accomplishments that I'm proud of
We achieved the right equilibrium in the images analysis without overfitting our model.
What I learned
Working with speech recognition and human interactions is hard because the number of possible outcomes increases exponentially with the number of steps.
What's next for Multisite Monitor
Provide a better UI and increase the images database so we can also diagnose plagues and other plants diseases.