Unnati | Microsoft Code.Fun.Do
Agri-Tech app using deep learning for disease prediction
We developed an android app to solve some of the most critical problems faced by farmers using disruptive solutions from the domain of computer vision. Plant disease prediction model trained using transfer learning over Inception-V3 pretrained on Imagenet. Model was deployed on an Azure server with a Flask based API. Interaction between the app and server carried out using HTTP requests. Our team was the 1st runner up in the campus round of the Microsoft Code.Fun.Do hackathon. Additionally, we also built a messenger chatbot in an effort to serve a wider audience without the hassle of installing the android app. DialogFlow was used for natural language interaction. We used the same Flask based API for serving our model to the chatbot.