Wireless sensor networks in agriculture through machine learning: A survey

•The survey paper consists of the concept of WSNs, Machine Learning (ML) and Deep Learning (DL).•The study of different agricultural issues.•ML techniques in WSN challenges.•Overcome the agricultural issues through WSN and ML concepts and techniques.•Deep Learning (DL) study for smart agriculture. This survey paper describes the concept of Wireless Sensor Networks (WSNs), Machine Learning (ML) and their applications on various fields of smart agriculture. Here we first define different types of WSNs which have direct impact on smart agriculture. The ML techniques are applied on data sets which are collected by sensor nodes deployed on the agricultural field. We also discuss the challenges or problems of the sensors networks and the ML techniques. We particularly focus on ML techniques and their uses to overcome the WSN’s challenges for agriculture. ML techniques are suitable for particular type (i.e. labelled/structured) of data sets to differentiate the given objects. Furthermore, we also consider the concept of Deep Learning technique for the general type of data sets that may be in vast amount. Finally, we discuss on various agricultural problems and previously used different techniques applied for smart or precision agriculture. Our main target is that the agricultural system to be smarter by using combination of different techniques, such as sensor networks, ML and deep learning. Finally this survey paper also gives the idea to use IoT and deep learning in agricultural WSNs for next level.