Suyin Zhou

Mrs. Suyin Zhou

orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 September 2023 in Plants
Verified
Journal Article
Open access
PDF file
TYTingting Yang
Tingting Yang
SZSuyin Zhou
Suyin Zhou
AXAijun Xu
Aijun Xu
JYJunhua YE
Junhua YE
JYJianxin Yin
Jianxin Yin
Accurate plant leaf image segmentation provides an effective basis for automatic leaf area estimation, species identification, and plant disease and pest monitoring. In this paper, based on our previous publicly available leaf dataset, an approach that fuses YOLOv8 and improved DeepLabv3+ is proposed for precise image segmentation of individual leaves. First, the leaf object detection algorithm-based YOLOv8 was introduced to reduce the interference of backgrounds on the second stage leaf segmentation task. Then, an improved DeepLabv3+ leaf segmentation method was proposed to more efficiently capture bar leaves and slender petioles. Densely connected atrous spatial pyramid pooling (DenseASPP) was used to replace the ASPP module, and the strip pooling (SP) strategy was simultaneously inserted, which enabled the backbone network to effectively capture long distance dependencies. The experimental results show that our proposed method, which combines YOLOv8 and the improved DeepLabv3+, achieves a 90.8% mean intersection over the union (mIoU) value for leaf segmentation on our public leaf dataset. When compared with the fully convolutional neural network (FCN), lite-reduced atrous spatial pyramid pooling (LR-ASPP), pyramid scene parsing network (PSPnet), U-Net, DeepLabv3, and DeepLabv3+, the proposed method improves the mIoU of leaves by 8.2, 8.4, 3.7, 4.6, 4.4, and 2.5 percentage points, respectively. Experimental results show that the performance of our method is significantly improved compared with the classical segmentation methods. The proposed method can thus effectively support the development of smart agroforestry.

15 September 2024 in Agriculture
Verified
Journal Article
Open access
PDF file
TYTingting Yang
Tingting Yang
SZSuyin Zhou
Suyin Zhou
AXAijun Xu
Aijun Xu
JYJunhua YE
Junhua YE
JYJianxin Yin
Jianxin Yin
In urban forest management, individual street tree segmentation is a fundamental method to obtain tree phenotypes, which is especially critical. Most existing tree image segmentation models have been evaluated on smaller datasets and lack experimental verification on larger, publicly available datasets. Therefore, this paper, based on a large, publicly available urban street tree dataset, proposes YOLO-SegNet for individual street tree segmentation. In the first stage of the street tree object detection task, the BiFormer attention mechanism was introduced into the YOLOv8 network to increase the contextual information extraction and improve the ability of the network to detect multiscale and multishaped targets. In the second-stage street tree segmentation task, the SegFormer network was proposed to obtain street tree edge information more efficiently. The experimental results indicate that our proposed YOLO-SegNet method, which combines YOLOv8+BiFormer and SegFormer, achieved a 92.0% mean intersection over union (mIoU), 95.9% mean pixel accuracy (mPA), and 97.4% accuracy on a large, publicly available urban street tree dataset. Compared with those of the fully convolutional neural network (FCN), lite-reduced atrous spatial pyramid pooling (LR-ASPP), pyramid scene parsing network (PSPNet), UNet, DeepLabv3+, and HRNet, the mIoUs of our YOLO-SegNet increased by 10.5, 9.7, 5.0, 6.8, 4.5, and 2.7 percentage points, respectively. The proposed method can effectively support smart agroforestry development.

16 October 2024 in Animals
Verified
Journal Article
Open access
PDF file
ZLZiyuan Liang
Ziyuan Liang
AXAijun Xu
Aijun Xu
JYJunhua YE
Junhua YE
SZSuyin Zhou
Suyin Zhou
XWxiaoxing weng
xiaoxing weng
SBSian Bao
Sian Bao
Continuous movement monitoring helps quickly identify pig abnormalities, enabling immediate action to enhance pig welfare. However, continuous and precise monitoring of daily pig movement on farms remains challenging. We present an approach to automatically and precisely monitor the movement of group-housed pigs. The instance segmentation model YOLOv8m-seg was applied to detect the presence of pigs. We then applied a spatial moment algorithm to quantitatively summarize each detected pig’s contour as a corresponding center point. The agglomerative clustering (AC) algorithm was subsequently used to gather the pig center points of a single frame into one point representing the group-housed pigs’ position, and the movement volume was obtained by calculating the displacements of the clustered group-housed pigs’ center points of consecutive frames. We employed the method to monitor the movement of group-housed pigs from April to July 2023; more than 1500 h of top-down pig videos were recorded by a surveillance camera. The F1 scores of the trained YOLOv8m-seg model during training were greater than 90% across most confidence levels, and the model achieved an mAP50-95 of 0.96. The AC algorithm performs with an average extraction time of less than 1 millisecond; this method can run efficiently on commodity hardware.

3 April 2025 in Animals
Verified
Journal Article
Open access
PDF file
ZYZhaoyang Yin
Zhaoyang Yin
泽王泽华 王
泽华 王
JYJunhua YE
Junhua YE
SZSuyin Zhou
Suyin Zhou
AXAijun Xu
Aijun Xu
Pig tracking contributes to the assessment of pig behaviour and health. However, pig tracking on real farms is very difficult. Owing to incomplete camera field of view (FOV), pigs frequently entering and exiting the camera FOV affect the tracking accuracy. To improve pig-tracking efficiency, we propose a pig-tracking method that is based on skeleton feature similarity, which we named GcnTrack. We used YOLOv7-Pose to extract pig skeleton key points and design a dual-tracking strategy. This strategy combines IOU matching and skeleton keypoint-based graph convolutional reidentification (Re-ID) algorithms to track pigs continuously, even when pigs return from outside the FOV. Three identical FOV sets of data that separately included long, medium, and short duration videos were used to test the model and verify its performance. The GcnTrack method achieved a Multiple Object Tracking Accuracy (MOTA) of 84.98% and an identification F1 Score (IDF1) of 82.22% for the first set of videos (short duration, 87 s to 220 s). The tracking precision was 74% for the second set of videos (medium duration, average 302 s). The pigs entered the scene 15.29 times on average, with an average of 6.28 identity switches (IDSs) per pig during the tracking experiments on the third batch set of videos (long duration, 14 min). In conclusion, our method contributes an accurate and reliable pig-tracking solution applied to scenarios with incomplete camera FOV.