Wei-zhen Liang

Dr. Wei-zhen Liang

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
31 January 2026 in Computers and Electronics in Agriculture
Verified
Journal Article
Open access
ZYZonglin Yang
Zonglin Yang
WLWei-zhen Liang
Wei-zhen Liang
NLNevin Lawrence
Nevin Lawrence
XQXin Qiao
Xin Qiao
BRBenjamin Riggan
Benjamin Riggan
DADiwenitissiou Philipine Andjawo
Diwenitissiou Philipine Andjawo

30 September 2025 in Preprints.org
Verified
Preprint
Open access
PDF file
Chi-En ChiangChi-En Chiang
Chi-En Chiang
WLWei-zhen Liang
Wei-zhen Liang
JCJingqiu Chen
Jingqiu Chen
XQXin Qiao
Xin Qiao
VTVioleta Tsolova
Violeta Tsolova
JOJoseph Oboamah
Joseph Oboamah
Bud break is a critical phenological stage in muscadine grapevines, marking the start of the growing season and the increasing need for irrigation management. Real-time bud detection enables irrigation to match muscadine grape phenology, conserving water and enhancing performance. This study presents BudCAM, a low-cost, solar-powered, edge-computing camera system based on Raspberry Pi 5 and integrated with LoRa radio board, developed for real-time bud detection. Nine BudCAMs were deployed at Florida A&M University Center for Viticulture and Samll Fruit Research from mid February to mid March, 2024, monitoring three wine cultivars (A-27, noble, and Floriana) with three replicates each. Muscadine grape canopy images were captured every 20 minutes between 7:00 to 19:00, generating 2656 high-resolution (4656×3456 pixels) bud break images as database for bud detection algorithm development. The dataset was divided into 70% training, 15% validation, and 15% test. YOLOv11 models were trained using two primary strategies: a direct single-stage detector on tiled raw images and a refined two-stage pipeline that first identifies the grapevine cordon. Extensive evaluation of multiple model configurations identified top performers for both the single-stage (mAP@0.5=86.0%) and two-stage (mAP@0.5=85.0%) approaches. Further analysis revealed that preserving image scale via tiling was superior to alternative inference strategies like resizing or slicing. Field evaluations during the 2025 growing season confirmed the system’s effectiveness, with the two-stage model showing greater robustness to environmental noise like lens fog. A time-series filter smooths the raw daily counts to reveal a clear phenological trend for visualization. In its final deployment, the autonomous BudCAM system captures an image, runs inference on-device, and transmits the bud count in under three minutes, demonstrating a complete, field-ready solution for precision vineyard management.

24 October 2025 in Agriculture
Verified
Journal Article
Open access
PDF file
Chi-En ChiangChi-En Chiang
Chi-En Chiang
WLWei-zhen Liang
Wei-zhen Liang
JCJingqiu Chen
Jingqiu Chen
XQXin Qiao
Xin Qiao
VTVioleta Tsolova
Violeta Tsolova
JOJoseph Oboamah
Joseph Oboamah
Bud break is a critical phenological stage in muscadine grapevines, marking the start of the growing season and the increasing need for irrigation management. Real-time bud detection enables irrigation to match muscadine grape phenology, conserving water and enhancing performance. This study presents BudCAM, a low-cost, solar-powered, edge computing camera system based on Raspberry Pi 5 and integrated with a LoRa radio board, developed for real-time bud detection. Nine BudCAMs were deployed at Florida A&M University Center for Viticulture and Small Fruit Research from mid-February to mid-March, 2024, monitoring three wine cultivars (A27, noble, and Floriana) with three replicates each. Muscadine grape canopy images were captured every 20 min between 7:00 and 19:00, generating 2656 high-resolution (4656 × 3456 pixels) bud break images as a database for bud detection algorithm development. The dataset was divided into 70% training, 15% validation, and 15% test. YOLOv11 models were trained using two primary strategies: a direct single-stage detector on tiled raw images and a refined two-stage pipeline that first identifies the grapevine cordon. Extensive evaluation of multiple model configurations identified the top performers for both the single-stage (mAP@0.5 = 86.0%) and two-stage (mAP@0.5 = 85.0%) approaches. Further analysis revealed that preserving image scale via tiling was superior to alternative inference strategies like resizing or slicing. Field evaluations conducted during the 2025 growing season demonstrated the system’s effectiveness, with the two-stage model exhibiting superior robustness against environmental interference, particularly lens fogging. A time-series filter smooths the raw daily counts to reveal clear phenological trends for visualization. In its final deployment, the autonomous BudCAM system captures an image, performs on-device inference, and transmits the bud count in under three minutes, demonstrating a complete, field-ready solution for precision vineyard management.

30 April 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
SXShuhua Xie
Shuhua Xie
AGAngie Gradiz
Angie Gradiz
XQXin Qiao
Xin Qiao
JOJoseph Oboamah
Joseph Oboamah
WLWei-zhen Liang
Wei-zhen Liang
PSPoNien Su
PoNien Su
Accurate estimation of daily crop evapotranspiration (ET) at the sub-field level is critical for precision irrigation scheduling and management, yet few studies have evaluated whether satellite-based ET products can resolve soil-driven spatial variability within a single field. We compared three ET estimation approaches across three management zones delineated by apparent electrical conductivity in a commercial dry edible bean field in western Nebraska, U.S.A. during the 2023 growing season, including Two-Source Energy Balance (TSEB) with ground-installed infrared radiometer thermometers (IRT) as the reference, soil water balance (SWB) from commercial capacitance sensors with factory default calibration, and satellite-based ET from OpenET. Three zones exhibited significant differences (p < 0.05) in available water, vegetation indices, and yield, confirming distinct growing environments under uniform irrigation. The OpenET ensemble showed moderate daily agreement with TSEB (R² = 0.33-0.52, RMSE = 1.0-1.2 mm/day), whereas SWB showed poor agreement (R² = 0.01-0.05, RMSE = 3.1-6.8 mm/day) due to sensor inaccuracies possibly caused by factory default calibration, preferential flow in sandy soils, and unquantifiable deep percolation. Among six individual OpenET models, eeMETRIC achieved the highest correlations (R² = 0.38-0.60) but failed to differentiate ET among zones, while SIMS and SSEBop combined relatively high correlations (R² = 0.35-0.58) with the ability to detect significant zone-level ET differences (p < 0.05). These findings highlight the limitations of capacitance soil moisture sensors for daily ET estimation under typical commercial conditions and demonstrate that select OpenET models can capture sub-field ET variability, supporting their integration into zone-level irrigation scheduling for spatially heterogeneous fields. By making zone-specific ET information available without on-farm instrumentation, these results help open a practical pathway toward broader producer adoption of data-driven precision irrigation scheduling.

1 January 2023
Verified
Conference Paper
Closed access
PDF file
JCJingqiu Chen
Jingqiu Chen
WLWei-zhen Liang
Wei-zhen Liang
JJJian Jin
Jian Jin
VTVioleta M. Tsolova
Violeta M. Tsolova

13 August 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
ZYZonglin Yang
Zonglin Yang
WLWei-zhen Liang
Wei-zhen Liang
NLNevin Lawrence
Nevin Lawrence
XQXin Qiao
Xin Qiao
BRBenjamin S. Riggan
Benjamin S. Riggan
FLFuchen Li
Fuchen Li
Precision weed control requires both species-level identification and per-instance localization, butconventional object detectors are constrained by a closed-vocabulary paradigm that hinders cross-regiondeployment and lack the reasoning capability to justify their predictions in complex agricultural scenes.Recently, multimodal large language models (MLLMs) have demonstrated strong visual perceptionand reasoning in visual grounding, offering a promising alternative. However, their insufficient domainspecificbotanical knowledge often leads to hallucinations during fine-grained weed identification. Inthis study,WeedExpert-R1 is introduced, a multimodal reasoning model that, following the R1-styletraining paradigm, incentivizes visually grounded botanical reasoning through verifiable rewards.Specifically, a domain-specific Chain-of-Thought (CoT) synthesis pipeline is proposed. It pairs ahuman-curated botanical trait dictionary, covering leaf shape, margin, petiole, and stem morphology,with an Auditor–Synthesizer LLM workflow to generate high-quality reasoning data for supervisedfine-tuning as a cold start. This pipeline bridges the gap between textual botanical knowledge and thevisual perception of MLLMs. Group Relative Policy Optimization (GRPO) is then applied withverifiable rewards (format, accuracy, count, and length penalty) to further enhance the model’scapability to perceive and localize diverse weeds. WeedExpert-R1-4B achieves 75.82%, 89.30%,and 87.81% on Precision@(F1=1, IoU≥0.5), Precision@0.5, and Recall@0.5 across 37 weed specieson a benchmark suite of six weed datasets (3SeasonWeedDet10, CottonWeedDet12, CottonWeedDet3,Weed-crop,Weed25, and PREEC), and outperforms both frontier proprietary models (GPT-5.4, Gemini-3.1-Pro) and larger open-source baselines (Qwen3-VL-30B-Instruct, Gemma-4-31B-it). Moreover,qualitative results on unseen species further illustrate the open-vocabulary capability ofWeedExpert-R1,suggesting potential for deployment across diverse regions and crops without retraining.