Agribot: A Natural Language Generative Neural Networks Engine for Agricultural Applications

In this paper, we present an artificially intelligent chatbot which would help farmers by providing solutions to various farming related problems and facilitate their decision-making process. The bot not only provides answers to frequently asked questions but also lays emphasis on crop disease detection and weather forecasting. We developed an end to end trainable sequence-to-sequence learning model with the objective of achieving conversational task-oriented system based on minimal assumption on its sequence structure. Our approach exploits a multilayered Long Short-Term Memory (LSTM) unit which maps the input sequence to a corresponding output sequence by converting it into a vector of fixed dimensionality in between. To achieve the disease detection, a Convolutional Neural Network architecture is implemented in which a multilayered architecture is developed and trained from scratch which would classify the plant images into different classes. For conversational system module we have used the Kisan Call Center (KCC) dataset which contains logs of calls at KCC by farmers whereas for disease detection module plant village dataset is used. After training 98% accuracy was achieved for conversational system module on training data and 94% for disease detection module on test data.

I. Introduction

Farmers face a lot of issues - be it how to combat specific crop diseases, fertilizers choices, pest infestations or unpredictable weather. There are KCC units to answer all these type of queries. But "They (KCC) don’t pick the call, ever". Also advice received on phone calls are hard to remember. Another option for farmer is to find the solution through google apps but the problem is "You ask one thing (to Google) and get thousand responses. What to do!" [1]. In order to overcome the above mentioned issue we have devised an interactive chatbot for farmers.