Cross-Domain Sensing · Multimodal Intelligence · Field-Validated
Real-world challenges rarely stay within single disciplines. Integrating optical imaging, dynamic systems control, and lightweight neural networks, we engineer end-to-end perception systems that translate fundamental research into measurable impact across agriculture, industry, and healthcare.
- Research domain playground
- Systems & Control
- Imaging & Vision
- Precision Measurement
- IT Integration
- Biomedical Signals
- Agri & Livestock
Validating technology under real-world operating conditions
Factory vibrations, thermal drift, varying light in barns and greenhouses, and biological variations present the toughest challenges for algorithms. By developing custom optical setups and multimodal sensors, we improve signal quality from the source, pairing hardware with lightweight models to ensure robust performance under real operating conditions.
Explore research areasAbout the lab
Who we are and what we do
NCHU | NCUT
The lab is jointly directed by Associate Professor Wen-Lin Chu (Department of Bio-industrial Mechatronics Engineering, National Chung Hsing University) and Professor Bo-Lin Jian (Department of Electrical Engineering, National Chin-Yi University of Technology). Our work spans image processing, signal processing, artificial intelligence (AI), agricultural and biological engineering, optical imaging, and machine-tool condition monitoring. We integrate imaging systems with multimodal sensing into systematic research and development, and turn academic results into working industrial applications.
Principal investigators
Wen-Lin ChuAssociate Professor · Dept. of Bio-industrial Mechatronics Engineering, National Chung Hsing University
Bo-Lin JianProfessor · Dept. of Electrical Engineering, National Chin-Yi University of Technology
Principles
What this lab runs on
01
Field-Tested Reliability
Research must go beyond simulation. We take our systems directly onto factory floors, into greenhouses, and across barns to test stability and accuracy under real operating conditions.
02
Empirical Integrity
We ground our work in empirical data. Field environments are full of noise; when results diverge from expectations, we investigate the underlying physics and refine the model rather than forcing the narrative.
03
Full-Stack Engineering
From optical benches and sensing circuits to signal processing and edge AI deployment, we train researchers who can independently engineer complete working prototypes.
04
Industry-Aligned Innovation
Driven by real bottlenecks across manufacturing, agriculture, and healthcare, we combine research grants and industry partnerships to turn technologies into actionable solutions.
Read the latest
Latest papers and updates
Page 1 / 1
Research domains
Research pillars
01 · Systems Engineering & Green Energy Control
Systems Engineering & Green Energy Control
Modelling and control of complex dynamic systems. Wind-turbine array wakes are measured on a precision schlieren bench and simulated for array monitoring and power optimisation, and the same control work extends to reinforcement-learning autonomous flight and CNC feedback compensation.
02 · Automated Imaging & Vision
Automated Imaging & Vision
Micro-scale defect inspection and flow-field visualisation. The work develops adaptive robust geometric fitting, automated inspection of night-vision optics, and real-time prediction from dynamic colour schlieren using deep neural networks.
03 · IT & System Integration
IT & System Integration
Deep-learning models are compressed for embedded deployment at the edge, delivering real-time anomaly alerts where compute is constrained, and combined with large language models (LLM) and retrieval-augmented generation (RAG) to build industrial knowledge platforms.
04 · Precision Measurement
Precision Measurement
Acquisition and analysis of weak signals under demanding conditions: terahertz single-pixel fast imaging (patented in Taiwan and the US), Z-type precision schlieren optical paths, image-based non-contact physiological measurement, and machine-tool thermal error compensation.
05 · Biomedical Signal Processing
Biomedical Signal Processing
EEG feature extraction, multimodal physiological signal analysis and visualisation, and objective assessment systems, validated through long-standing collaborations with clinical psychiatry, dementia care and horticultural therapy teams.
06 · Agricultural Technology & Smart Livestock
Agricultural Technology & Smart Livestock
Image-based plant disease recognition, field machinery automation, precision growth analysis, and real-time monitoring of dairy cattle lameness and poultry health, with every result validated on working farms and in greenhouses.
Impact
Research results
Precision livestock
Intelligent Cattle Lameness Monitoring
A lightweight MobileNetV3 lameness classifier fuses gait energy images (GEI) and history energy images (HEI) to grade lameness severity automatically.
Depth sensing
Non-contact Weight Prediction for Holstein Cows
Depth images of the back, rump and flank captured with an Intel RealSense D455 feed a processing pipeline that extracts body-surface features to predict weight change in non-pregnant and postpartum stages.
Thermal imaging
Heat Stress Analysis in Dairy Cattle
Muzzle temperature and respiration rate are extracted automatically from thermal image sequences to classify heat stress severity. Non-contact monitoring avoids the stress that measurement itself would cause, so farms can cool the herd earlier.
Join Us & Academic Community
Academic community
We serve as journal reviewers, guest editors and conference programme committee members, and accept speaking invitations from departments and industry associations. Write to us directly about collaboration or data exchange.
See academic serviceJoin the lab
We welcome master's, doctoral and undergraduate project students with an interest in computer vision, optical inspection, biomedical signals, smart manufacturing or precision agriculture. You will join active industry projects and research grants, and build a system end to end, from the optical bench to algorithm deployment.
Prospective master's, doctoral and undergraduate project students are welcome to get in touch.
Meet the team