Segmentation of Unsound Wheat Kernels Based on Improved Mask RCNN

Abstract

The grade of wheat quality depends on the proportion of unsound kernels. Therefore, the rapid detection of unsound wheat kernels is important for wheat rating and evaluation. However, in practice, unsound kernels are hand-picked, which makes the process time-consuming and inefficient. Meanwhile, methods based on traditional image processing cannot divide adherent particles well. To solve the above problems, this paper proposed an unsound wheat kernel recognition algorithm based on an improved mask RCNN. First, we changed the feature pyramid network (FPN) to a bottom-up pyramid network to strengthen the low-level information. Then, an attention mechanism (AM) module was added between the feature extraction network and the pyramid network to improve the detection accuracy for small targets. Finally, the regional proposal network (RPN) was optimized to improve the prediction performance. Experiments showed that the improved mask RCNN algorithm could identify the unsound kernels more quickly and accurately while handling adhesion problems well. The precision and recall were 86% and 91%, respectively, and the inference time on the test set with about 200 targets for each image was 7.83 s. Additionally, we compared the improved model with other existing segmentation models, and experiments showed that our model achieved higher accuracy and performance than the other models, laying the foundation for wheat grading.

Non-destructive detection of single-seed viability in maize using hyperspectral imaging technology and multi-scale 3D convolutional neural network

“…RGB images of maize seeds were captured using BASLER industrial cameras (acA1920-25um/uc, BASLER AG, Germany, 2.4 MP,100 fps) during germination test ( Figure 2D ) ( Shen et al., 2023 ). An adjustable camera platform was built to ensure consistency of the images and prevent camera shake.…”

“…(2021) used the Mask R-CNN algorithm to effectively segment and measure leaf characteristics and obtained an error rate of around 5%. An enhanced algorithm based on the mask RCNN was introduced by Shen et al. (2023) to recognize defective wheat kernels.…”

A Rapid Construction Method for High-Throughput Wheat Grain Instance Segmentation Dataset Using High-Resolution Images

“…Both the object detection and segment model components of the pipeline are interchangeable plugins. In future work, we can further compare various state-of-the-art grain instance segmentation methods [25][26][27] and replace the detection component of the pipeline.…”

A density map-based method for counting wheat ears

“…These studies have made significant progress toward accurate and efficient wheat ears segmentation, paving the way for improved agricultural research and crop yield estimation. To effectively deal with adhesion, ( Shen et al., 2023a ) proposes an improved Mask R-CNN-based algorithm for unsound wheat kernel segmentation, which achieves faster and more accurate unsound kernel recognition by means of a bottom-up feature pyramid network and by adding an Attention Mechanism (AM) module between the feature extraction network and the pyramid network, and at the same time effectively handles the sticking problem to achieve an accuracy of 86% and a recall of 91%. The inference time of this model on the test set is 7.83 seconds, which is significantly better than other segmentation models and provides an important foundation for wheat grading.…”