OVERVIEW OF RECENT DEEP LEARNING METHODS APPLIED IN FRUIT COUNTING FOR YIELD ESTIMATION

Abstract

This paper is an overview of the latest advancements of image recognition for fruit counting and yield estimation. Considering this domain is developing rapidly, we have considered the cutting-edge literature in the field, for the last 5 years, focused on the task of yield estimation by detecting and counting fruit in the tree canopy. This is a much more complex task than the classification of fruit post-harvesting, which has been more widely reviewed. Moreover, we identify the major challenges and propose the next steps for advancing this research field.

Keywords

• smart-agriculture, deep learning, yield estimation, transfer learning, intersection over union, F1-score.