A Review of Pre-processing Techniques for Weed-Plant Detection and Classification in Precision Agriculture

Precision agriculture has recently gained significant importance in computer vision technologies. Various processes as a part of agricultural production cycle from planting to harvesting can be carried out automatically and effectively by using computer vision. The lack of publicly available image datasets is a major obstacle to the rapid design and assessment of computer vision based applications and also to machine learning algorithms which support these applications. To reduce this bottleneck, numerous image dataset collections have been discovered and made publicly available since 2015. In spite of this development, there is still a need to focus on survey of these datasets. Two most important concerns—choosing the right dataset and knowing how to pre-process and prepare the images in datasets, are considerably challenging task in every application. This review paper gives a thorough analysis of the public image datasets and numerous pre-processing techniques carried out in the field of precision agriculture. This thorough study can lead to development of suitable methods for improved quality and productivity of the crop along with proper weed management.