Machine learning approaches are progressively successful in image based

analysis such as different diseases prediction as well as level of risk

assessment. In this paper, image based data analysis with machine learning

technique was applied to fertilizer treatment of maize. We address this issue

as our country depend on agricultural field rather than others. Maize has a

bright future. To predict fertilizer treatment of maize dataset was comprised

of ground coverage region which highlights the green pixels of a maize

image. For calculating green pixels from an image we used “Can Eye” tool.

The achievement of machine learning approaches is highly dependent on

quality and quantity of the dataset which is used for training the machine for

better classification result. For this perseverance, we have collected images

from the maize field directly. Then processed those images and classified the

data into four classes (Less Nitrogen=-N, Less Phosphorus=-P,

Less Potassium=-K and NPK) to train our machine using decision tree

algorithm to predict fertilizer treatment. We have got 93% classification

accuracy for decision tree. Finally, the outcome of this paper is fertilizer

treatment of a maize field based on the ground cover percentage, and we

implemented this whole paper work using an android platform because of the

availability of android mobile phone throughout the world