RESUMO
The total boll count from a plant is one of the most important phenotypic traits for cotton breeding and is also an important factor for growers to estimate the final yield. With the recent advances in deep learning, many supervised learning approaches have been implemented to perform phenotypic trait measurement from images for various crops, but few studies have been conducted to count cotton bolls from field images. Supervised learning models require a vast number of annotated images for training, which has become a bottleneck for machine learning model development. The goal of this study is to develop both fully supervised and weakly supervised deep learning models to segment and count cotton bolls from proximal imagery. A total of 290 RGB images of cotton plants from both potted (indoor and outdoor) and in-field settings were taken by consumer-grade cameras and the raw images were divided into 4350 image tiles for further model training and testing. Two supervised models (Mask R-CNN and S-Count) and two weakly supervised approaches (WS-Count and CountSeg) were compared in terms of boll count accuracy and annotation costs. The results revealed that the weakly supervised counting approaches performed well with RMSE values of 1.826 and 1.284 for WS-Count and CountSeg, respectively, whereas the fully supervised models achieve RMSE values of 1.181 and 1.175 for S-Count and Mask R-CNN, respectively, when the number of bolls in an image patch is less than 10. In terms of data annotation costs, the weakly supervised approaches were at least 10 times more cost efficient than the supervised approach for boll counting. In the future, the deep learning models developed in this study can be extended to other plant organs, such as main stalks, nodes, and primary and secondary branches. Both the supervised and weakly supervised deep learning models for boll counting with low-cost RGB images can be used by cotton breeders, physiologists, and growers alike to improve crop breeding and yield estimation.
Assuntos
Aprendizado Profundo , Gossypium , Melhoramento VegetalRESUMO
The Genetically Modified (GMO) Corn Experiment was performed to test the hypothesis that wild animals prefer Non-GMO corn and avoid eating GMO corn, which resulted in the collection of complex image data of consumed corn ears. This study develops a deep learning-based image processing pipeline that aims to estimate the consumption of corn by identifying corn and its bare cob from these images, which will aid in testing the hypothesis in the GMO Corn Experiment. Ablation uses mask regional convolutional neural network (Mask R-CNN) for instance segmentation. Based on image data annotation, two approaches for segmentation were discussed: identifying whole corn ears and bare cob parts with and without corn kernels. The Mask R-CNN model was trained for both approaches and segmentation results were compared. Out of the two, the latter approach, i.e., without the kernel, was chosen to estimate the corn consumption because of its superior segmentation performance and estimation accuracy. Ablation experiments were performed with the latter approach to obtain the best model with the available data. The estimation results of these models were included and compared with manually labeled test data with R 2 = 0.99 which showed that use of the Mask R-CNN model to estimate corn consumption provides highly accurate results, thus, allowing it to be used further on all collected data and help test the hypothesis of the GMO Corn Experiment. These approaches may also be applied to other plant phenotyping tasks (e.g., yield estimation and plant stress quantification) that require instance segmentation.