MCFF: Plant leaf detection based on multi-scale CNN feature fusion

Abstract

Plant leaf detection is one of the essential aspects of the scientific plant breeding and precision agriculture process. Manual detection requires professional knowledge of the operators, high labor costs, and long time-consuming cycles. To this end, this paper proposes a multi-scale CNN feature fusion (MCFF) to detect the Rosette plant, Arabidopsis, and Tobacco. The experimental results indicate that the mean average precision of the proposed method is higher than the traditional methods such as RetinaNet, CenterNet, and Faster R-CNN.