A Method of Crop Seedling Plant Segmentation on Edge Information Fusion Model

Abstract

Automatic segmentation of plant images is a hot issue in plant phenotyping research. It is also one of the core technologies for applications such as crop growth process monitoring and pest identification. Due to the different scales and sizes of fruits, branches and leaves of fruit and vegetable plants in the natural environment, and irregular edges, it is difficult to accurately segment. In order to accurately segment crop seedlings in natural environment and realize automatic measurement of seedling location and phenotype, this paper proposes a crop seedling plant segmentation network model that fuses the semantic and edge information of target regions. The backbone network is composed of the UNET network, which guides the backbone network to perceive the plant edge information when extracting features; uses the spatial hole feature pyramid to build a feature fusion module, which fuses the features extracted by the UNET backbone network and the edge perception module. Combining edge-aware loss and feature fusion loss, a joint loss function is constructed for overall network optimization. The encoderdecoder network is referenced in the study. The encoding network uses densenet to reuse and fuse multilayer features to improve the way of information transmission; the decoding network uses transposed convolution for upsampling, combined with layer jump connections to fuse shallow detail information and deep semantic information; add a hole between encoding and decoding Atrous spatial pyramid pooling (ASPP) to extract feature maps of different receptive fields to integrate multi-scale features and aggregate contextual information. The experimental results show that under the same network training parameters, the average cross-merging rate and average recall rate obtained by testing the method in this paper are 58.13% and 64.72%, respectively, which are better than the segmentation results corresponding to the manually labeled samples; in addition, adding in the training samples After 10% of the outdoor seedling images, the average pixel accuracy of the proposed method on the outdoor test set can reach 90.54%, with good generalization ability.

Extracting Citrus-Growing Regions by Multiscale UNet Using Sentinel-2 Satellite Imagery

“…The proposed method has the structure of detail learning that helps differentiate other crops. The idea of combining this model with the ASPP module is inspired by [51,52], and experiments confirm that this combination is effective. The comparison shown in Figure 8f,g demonstrates that the proposed model can distinguish the borders of citrus-growing regions.…”

Clasificación de calidad de manzana para monitoreo de cosechabilidad utilizando visión por computador y algoritmos de aprendizaje profundo

“…Este estudio permite extraer rasgos del fruto para el monitoreo y evaluación de la cosecha. Un mayor avance se encuentra en la clasificación del tipo de madurez de manzana utilizando imágenes RGB bajo una arquitectura modificada En-UNet [36], alcanzando una precisión del 97,54% en la validación del modelo de segmentación. Sin embargo, dichos porcentaje se reducen drásticamente a menos del 93,1% cuando el nivel intermedio de madurez de la fruta no está bien definido.…”

“…La configuración de la red, y programación de los algoritmos fueron implementados utilizando librerías de PyTorch. SSD-Mobilenet [36] es una arquitectura de red para la detección de objetos, donde una de sus características fundamentales es su funcionamiento en tiempo real en dispositivos móviles e integrados. Además, esta red combina el detector SSD-300 Single-Shot MultiBox [37] con una red troncal MobileNet [36], como se observa en la Figura 1.…”

Testing a Method Based on an Improved UNet and Skeleton Thinning Algorithm to Obtain Branch Phenotypes of Tall and Valuable Trees Using Abies beshanzuensis as the Research Sample

“…UNet [21] was initially applied in medical image segmentation as an early CNN because UNet requires only a small amount of data to produce accurate segmentation results. UNet has also been applied in plant segmentation tasks by several researchers, who have proposed many enhancements to the UNet network structures to improve the semantic segmentation performance [22,23]. In addition, attention mechanisms have shown outstanding performance in the natural language processing (NLP) and computer vision tasks of the commonly used models.…”