Seed Dispenser using Drones and Deep Learning Techniques for Reforestation

As known already, Deforestation still continues to degrade the earth at an alarming rate, where it destroys the surrounding flora and fauna. According to the Food and Agriculture Organization of the United Nations it was estimated that between 2016 and 2020, the area of forest reduced from, 16 million hectares per year to 10 million hectares per year. The main and primary reason for this, is continuing to be the agricultural expansion and commercial production, which leads to loss of forest biodiversity. Almost 40 percent of the degradation is accounted to this factor. Looking into the bright side of forest it would bring around 86 million jobs, which would support the livelihood of many. In the world roughly around 870 million people spend time collecting wood for fuel (fuel-wood) to produce charcoal.One of the major effects caused by deforestation is the greenhouse effect, here the main gases namely CO2 (carbon-dioxide), CH4 (methane), NO2 (nitrogen dioxide) and cholofloro carbons. These gases trap the sunlight which fall in the IR range, this disturbs the equilibrium of the earth. In addition to this cutting trees also release CO2, adding to the rigid hold of these gases on the earth, which ultimately lead to the warming of the earth, this is very dangerous for humans and many lifeforms on earth.In this paper a novel solution to counter the problem of deforestation using drones and deep learning techniques has been proposed. The main reason of considering drones for this purpose is that it is fast as compared to traditional manual labor (almost 9 times faster), which saves time and improves efficiency. Moreover, a drone being unmanned avoids any risk of human danger, this method can be best suited for area effected by natural disaster like a forest fire in remote area, where people cannot travel, make reforestation at those areas relatively easy. Also, this method can be extended towards monitoring of the number of trees in a locality making it easier to keep tracks of the areas of interest. Moreover, with an expert steering the operations of seed dispersion, appropriate areas of sowing seeds can be found in different topologies with an advantage of object detection at these areas. Here a Mobile Net architecture is proposed and implemented for object detection using raspberry pi, which is embedded in the drone along with a seed dispenser, which will be used to sow the seeds at the barren or "deforested land".