Скачать или смотреть GreenEYE | Classification of Aerial Images using Hybrid Machine-Vision Learning

The cycle of agricultural production must include crop monitoring. It is essential for disease detection, phenotyping, yield calculation, and weed management, all of which have a significant influence on the environment and national economies (Hayes and Decker 1996). Crop monitoring mistakes may lead to the loss of priceless resources like water and fertiliser.Traditionally, crop monitoring involved human labour since the field manager or landowner had to personally watch the crops. This takes time and is prone to human mistake (Li et al. 2019). Precision agriculture has helped to reduce human labour on the fields over the last decade by delivering crop monitoring technologies that are lesssubjective, cost-effective, and resilient. With recent technological breakthroughs, it is now possible to construct automated and non-destructive remote sensing-based techniques. The rising availability of unmanned aerial aircraft (UAV) is a promising option for remotely and flexible data collecting on agricultural fields without the physical labour that is normally required (Rokhmana 2015; Sarron et al. 2018). Crop growers do not have to manually survey plots but may instead use a UAV to take aerial images of their crops, which can then be analysed to gather information at the plant level. The advantages of UAVs include their commercial availability and the fact that most of them can now be flown with an autopilot. Furthermore, UAVs may fly at lower altitudes with greater safety and cheaper cost than human aircraft while obtaining higher spatial resolutions. Those obtained using a UAV can also attain higher spatial resolution than images given by satellite services and can cover more than a few hundred hectares per day under favourable weather circumstances (Rokhmana 2015). UAVs are already being utilised in precision agriculture to increase profitability and production by supplying farmers with synoptic data and task maps (Tokekar et al. 2016). While employing UAVs can be a cheaper and faster way to collect aerial data, this data collection provides little value unless the received photographs are translated into usable information. Aerial images provide little value to the farmer until they are translated into operational knowledge. Machine vision technologies may extract significant information from high-resolution photos. The goal of this study is to create a fully automated system for calculating the number of plants in a field using very-high resolution RGB UAV footageto analyse crop emergence. The research topic posed in this paper was if it was possible to count the number of plants using very-high resolution RGB UAV footage. The suggested algorithm is a hybrid machine vision-learning technique.