Although season-long cotton flower counts have value to breeders and growers, a manual data collection process is too laborious to be practical in most cases. In recent years, several fully automated flower counting approaches have been proposed. However, such approaches are typically designed to run offline and require a significant amount of computation. Furthermore, little thought has gone into developing convenient interfaces and integrations so that a layperson can use such systems without extensive training. The goal of this study is to develop a lightweight flower tracking system that is deployable on a ground robot and can operate in real-time. We modify a previous GCNNMatch++ approach to increase the inference speed. Additionally, we fuse data from multiple cameras in order to avoid canopy occlusions, and extract three-dimensional flower locations by integrating GPS data from the robot. We show that our approach significantly outperforms UAV-based counting and single-camera counting while running at above 40 FPS on an edge device, achieving a counting error of 15% and an average localization error of 19 cm. This level of performance is enough to observe significant differences in flowering behavior between genotypes. Overall, we believe that our highly-integrated, automated, and simplified flower counting solution makes significant strides towards a practical commercial cotton phenotyping platform.