Wenjing Li

Wenjing Li

Auto-generated profile, based on publicly available information

Wenjing Li

Inner Mongolia University of Technology

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

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

Representative Publications

View All Publications
18 June 2022 in Mobile Information Systems
Verified
Journal Article
Open access
PDF file
WLWenjing Li
Wenjing Li
ZLZhiQiang Liu
ZhiQiang Liu
ZHZhipeng Hu
Zhipeng Hu
As one of the staple foods recognized by everyone, the quality of the potato determines the popularity of the potato. In order to grow healthy potatoes and improve the quality of potatoes, this research is based on multimodal sensor data fusion technology to achieve data detection and unified management. And it can analyze the data to find out the relationship between the application of nitrogen and potassium fertilizers and potato growth and quality, so as to have a reasonable plan for potato cultivation. This paper firstly described the research results of multimodal sensors, data fusion, nitrogen and potassium fertilizers, and potato quality, and then briefly introduced the multi-sensor data fusion technology, mainly including application areas, methods, and structures. Finally, the effects of nitrogen and potassium fertilizers on the growth and quality of potato were explained through the analysis of experimental data. And the experimental results showed that under different potassium treatments, Dongnong 311 had the highest starch content under K3 treatment, which was 13.9%. This study provides a certain reference for the application of nitrogen and potassium fertilizers in potato cultivation.

10 March 2026 in IEEE Access
Verified
Journal Article
Open access
JGJia Guo
Jia Guo
ZLZhiqiang Liu
Zhiqiang Liu
XZXu Zhang
Xu Zhang
RWRuili Wang
Ruili Wang
XMXiao Ma
Xiao Ma
WLWenjing Li
Wenjing Li
Automated tomato detection is fundamental for intelligent management in plant factories, where high-value crops are cultivated under fully controlled environments dominated by artificial LED lighting. Compared with conventional greenhouses or open fields, plant factories exhibit severe spectral shifts, non-uniform illumination with strong highlights and deep shadows, and dense occlusions caused by high-density planting and structural components. These factors make it difficult for existing detectors to simultaneously maintain robustness, accuracy, and efficiency on resource-constrained edge devices. To address the coupled challenges of illumination-aware feature extraction, multi-scale dense-target detection, and lightweight deployment, we propose SlimTom-DETR, a real-time tomato detection model built upon the RT-DETR- R18 baseline. SlimTom-DETR integrates three tightly coupled components: a C2f_DMB dynamic mixing backbone that adaptively adjusts receptive fields via multi-branch depthwise separable convolutions and gradient-diversion connections, an ELA-HSFPN feature fusion network that embeds efficient local attention into a hierarchical feature pyramid to enhance spatial selectivity and cross-scale alignment for dense small fruits, and a Polarized Self-Attention Decoupling Module (PSDM) that replaces conventional multihead attention with polarization-aware linear attention and a spatially enhanced feedforward network to explicitly decouple highlight and shadowresponses. Experiments on the Tomato-Plant-Factory-Dataset show that SlimTom-DETR achieves 90.8% mAP@50 and 69.5% mAP@50:95, surpassing RT-DETR-R18 by 2.0 and 2.2 percentage points, respectively, while reducing GFLOPs by 27.4% (56.9 to 41.3) and parameters by 20.2% (19.8M to 15.8M). Comprehensive comparisons, ablation studies, and cross-domain evaluations on Laboro Tomato,WeedCrop, and VisDrone further verify that SlimTom-DETR attains a favorable accuracy–efficiency trade-off and strong generalization ability in complex agricultural environments.