Abstract:In complex environments, overlapping leaves and uneven light can make pixels of leaf edges difficult to identify, resulting in a poor segmentation performance of the target leaf. In addition, the pixel ratio imbalance between the background area and the target area is the main reason that undermines the accuracy of spot extraction. To address these problems, a novel two-stage DeepLabv3+ with adaptive loss is proposed for the segmentation of apple leaf disease images in complex scenes. The proposed adaptive loss adds a modulation factor to the cross-entropy (CE) loss that can reduce the weight of losses generated by easily classified pixels. Therefore, it allows the model to focus more on hard-to-classify pixels during learning, thus improving segmentation accuracy. The novel two-stage model, consisting of Leaf-DeepLabv3+ and Disease-DeepLabv3+, is named LD-DeepLabv3+. In the first stage of the proposed model, Leaf-DeepLabv3+ is employed to extract the leaves from the complex environment. At this stage, the receptive field block (RFB) and the reverse attention (RA) module are introduced to improve the perception ability of the model for different sizes of blades and their edges. Then, the Disease-DeepLabv3+ is designed to segment disease spots from the erased background leaf images in the second stage of the proposed model. In the Disease-DeepLabv3+, the rates of the dilated convolution in atrous spatial pyramid pooling (ASPP) are adjusted to make it more suitable for extracting smaller targets, and the channel attention block (CAB) is introduced to highlight significant spot information and suppress unimportant information. The experimental results show that the proposed method, which combines LD-DeepLabv3+ with the adaptive loss, reaches 98.70% intersection over union (IoU) for leaf segmentation and 86.56% IoU for spot extraction. Compared with the two-stage model DUNet, the proposed method improves the segmentation accuracy of leaves and spots by 0.93% and 4.27%, respectively. Moreover, the total number of parameters and floating points of operations of the proposed method are only 16.96% and 18.25% of those of DUNet, respectively. Hence, the proposed method can provide an effective solution to extract leaves and disease spots in complex environments and has lower computational costs. This makes it suitable for deployment on mobile devices for applications in agriculture.

AC-UNet: an improved UNet-based method for stem and leaf segmentation in Betula luminifera

An Improved Res-UNet Model for Tree Species Classification Using Airborne High-Resolution Images

An Improved U-Net-based in Situ Root System Phenotype Segmentation Method for Plants

An Approach for Plant Leaf Image Segmentation Based on YOLOV8 and the Improved DEEPLABV3

PlantNet: A dual-function point cloud segmentation network for multiple plant species

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

MFBP-UNet: A Network for Pear Leaf Disease Segmentation in Natural Agricultural Environments

RAAWC-UNet: an apple leaf and disease segmentation method based on residual attention and atrous spatial pyramid pooling improved UNet with weight compression loss

Stem–Leaf Segmentation and Phenotypic Trait Extraction of Individual Plant Using a Precise and Efficient Point Cloud Segmentation Network

Semantic segmentation of remote sensing images combined with attention mechanism and feature enhancement U-Net

A novel approach for apple leaf disease image segmentation in complex scenes based on two-stage DeepLabv3+ with adaptive loss

ACE R-CNN: An Attention Complementary and Edge Detection-Based Instance Segmentation Algorithm for Individual Tree Species Identification Using UAV RGB Images and LiDAR Data

An effective image-based tomato leaf disease segmentation method using MC-UNet

CM-Unet: A Novel Remote Sensing Image Segmentation Method Based on Improved U-Net

The improved stratified transformer for organ segmentation of Arabidopsis

LeafNet: a Tool for Segmenting and Quantifying Stomata and Pavement Cells.

TreeUNet: Adaptive Tree Convolutional Neural Networks for Subdecimeter Aerial Image Segmentation

Cucumber Leaf Segmentation Based on Bilayer Convolutional Network

Straw Segmentation Algorithm Based on Modified UNet in Complex Farmland Environment

LIT-Unet: a lightweight and effective model for medical image segmentation

FG-UNet: fine-grained feature-guided UNet for segmentation of weeds and crops in UAV images