Abstract:
Human existence cannot be envisioned without agriculture. Soil is one of the major components
of agriculture. Depending on the PH scale, soil temperature and moisture of soil, various types of
soil are suitable for a variety of crops and food grains. But due to the farmer’s lack of absolute
knowledge about soil utility makes them face many more difficulties in the way of growing
crops. For these inconveniences, every year a huge number of crops are being wasted.
Furthermore, crops are also being damaged by various types of diseases and lack of quick
remedy adopted by the farmers. The fastest stratagem of predicting plant diseases is to analyze
leaf’s physiognomy changes and compare them with their actual color, shape, structure, etc. We
have used Convolutional Neural Network as a training method. CNN works via 3 dimensions of
layers where neurons of every layer aren’t fully connected to the next layer rather only a small
portion is connected and the output will be decreased to a single dimension. For this, even with
big datasets CNN works faster than any other networks. The program will exert plant images as
input and detaching them to predict plant diseases. Plant disease recognition on the basis of leaf’s
physiognomy changes and embedded based agriculture system are the fundamental purpose of
our project. This paper represents a system where it is possible to predict and suggest which
crops are compatible with any specific lands based on soil moisture and PH scale and all the
information related to the weather in a particular area and detect the actual diseases of different
types of crops.
