Image Analysis Based on Heterogeneous Architectures for Precision Agriculture: A Systematic Literature Review

Precision agriculture (AP) is a management strategy that uses ICT (Information and Communication Technologies) to obtain information from different sources in order to support decision-making, considering environmental and economic aspects to optimize the Farmer’s tasks and provide quality products to the costumer. The application of AP in agriculture can reduce time spent in manual activities, avoid the indiscriminate use of chemicals, increase production costs, soil deterioration and environmental pollution. Nowadays, AP is a booming area that, taking advantage of technological advances, in computer vision, heterogeneous architectures (Multicore, GPU, FGPA) and artificial intelligence techniques (Machine learning, Deep learning), has allowed to systematize a variety of agricultural activities, such as disease detection, plant counting, and identification of weed, pests and insects in different crops. This paper presents a systematic review of literature (SRL) of image analysis and processing techniques applied in precision agriculture using heterogeneous technologies. Therefore, 32 scientific articles of the last five years from four relevant bibliographic databases (Scopus, ScienceDirect, IEEE Xplore, SpringerLink) were analyzed and synthesized. The selected publications answer to four research questions proposed in this study. From the obtained results, great opportunities for image analysis (segmentation), machine learning and the use of graphic accelerators (GPU) were identified, which stand out as promising techniques and tools for the development of efficient and precise automatic systems, with the perspective to its application in real time for many agricultural tasks.