Automatic rape flower cluster counting method based on low-cost labelling and UAV-RGB images

Abstract

Background The flowering period is a critical time for the growth of rape plants. Counting rape flower clusters can help farmers to predict the yield information of the corresponding rape fields. However, counting in-field is a time-consuming and labor-intensive task. To address this, we explored a deep learning counting method based on unmanned aircraft vehicle (UAV). The proposed method developed the in-field counting of rape flower clusters as a density estimation problem. It is different from the object detection method of counting the bounding boxes. The crucial step of the density map estimation using deep learning is to train a deep neural network that maps from an input image to the corresponding annotated density map. Results We explored a rape flower cluster counting network series: RapeNet and RapeNet+. A rectangular box labeling-based rape flower clusters dataset (RFRB) and a centroid labeling-based rape flower clusters dataset (RFCP) were used for network model training. To verify the performance of RapeNet series, the paper compares the counting result with the real values of manual annotation. The average accuracy (Acc), relative root mean square error (rrMSE) and $$R^2$$ R 2 of the metrics are up to 0.9062, 12.03 and 0.9635 on the dataset RFRB, and 0.9538, 5.61 and 0.9826 on the dataset RFCP, respectively. The resolution has little influence for the proposed model. In addition, the visualization results have some interpretability. Conclusions Extensive experimental results demonstrate that the RapeNet series outperforms other state-of-the-art counting approaches. The proposed method provides an important technical support for the crop counting statistics of rape flower clusters in field.

Morning Glory Flower Detection in Aerial Images Using Semi-Supervised Segmentation with Gaussian Mixture Models

“…This characteristic enhances the model's adaptability and practicality, positioning it as a valuable solution for flower detection in aerial images while minimizing the need for extensive annotated datasets. By incorporating user interaction and semi-supervised learning into our enhanced GMM framework, we aim to generate accurate ground truths for future deep learning models [47,48]. Apart from I. purpurea flower counting, our method also provides valuable segmentation results (Figure 6).…”

A method for small-sized wheat seedlings detection: from annotation mode to model construction

“…However, high-quality annotated data has always been crucial in constructing and applying object detection models [ 10 – 12 ]. In previous studies, annotation patterns have been optimized by setting the annotated regions’ size and adjusting the bounding boxes’ orientation to improve the acquisition of annotated data [ 13 – 15 ]. However, individual wheat seedlings are tiny and show significant image morphological variations.…”

“…In addition, the interference of the soil background is significant, resulting in low detection efficiency of the model. Some researchers have proposed alternative annotations of key parts, such as leaf tips and local, instead of annotating the whole plant [ 15 ]. However, due to the mechanical or drill sowing for wheat, the seedlings have small local sizes and dense distributions during the seedling stage [ 16 ].…”

Accurate and fast implementation of soybean pod counting and localization from high-resolution image

“…With the widespread availability of low-cost digital cameras and high-performance graphics processing units, deep learning methods have enabled the shift from traditional manual labor to automated solutions for plant vision application. Various methods have been developed for different applications, such as estimating ear density (Xiong et al, 2019), locating cotton balls (Sun et al, 2022), detecting maize tassels , and counting rape flower clusters (Li et al, 2023). Lu et al (2023) proposed the YOLOv8-UAV model, which introduces a simple and effective up-sampling process.…”