Joseph Smith

👨‍🎓 About Me

Machine Learning Engineer and Computer Vision Researcher with 5+ years of experience developing and deploying advanced neural networks for image and video understanding, synthetic data generation, and robustness-critical visual systems. Strong background in PyTorch-based R&D, generative and augmentation-driven pipelines, and large-scale training and evaluation of deep neural networks.

Proven track record of state-of-the-art research published at CVPR and ECCV (Best Paper Award), with hands-on experience spanning graphics-aware data generation (Unreal Engine 5), domain adaptation, and synthetic-to-real generalisation. Experienced in collaborating across multidisciplinary R&D teams and translating research prototypes into production-ready systems for real-world deployment.

🎓 Education

• PhD in Computer Science

  Newcastle University, Autumn 2021 - December 2025

  Thesis: Robust and Explainable Deep Learning for Fine-Grained Classification in Brand Protection

  Industrial collaboration with P&G; research integrated into production systems.

• BSc in Computer Science, First Class Honours

  Newcastle University, Autumn 2018 - Summer 2021

  Dissertation: Fine-Grained Image Classification Using Siamese Neural Networks and Detection of Unseen Classes

💼 Work History

• Visiting Scientist

  German Aerospace Center (DLR), Sankt Augustin, April 2025 - July 2025

  Built an end-to-end drone detection pipeline using YOLOv5 / PyTorch, improving synthetic-to-real robustness.

  Generated large-scale synthetic datasets in Unreal Engine 5 & AirSim across varied weather domains and viewpoints.

  Created automation tools for dataset generation, reducing manual workload and improving reproducibility.

  Contributed to domain shift and synthetic data research; pipelines adopted for ongoing internal use.

• PhD Researcher (with Procter & Gamble)

  Newcastle University, 2021-2025

  Developed fine-grained counterfeit detection systems using thousands of real and synthetic samples.

  Designed re-adaptation and augmentation frameworks outperforming internal baselines under domain shift.

  Evaluated robustness under blur, noise, compression, and sensor artefacts; influenced P&G inspection workflows.

  Published at CVPR, ECCV, VISAPP; received Best Paper Award.

  Research deployed in production pipelines for model retraining and quality control.

• Class Demonstrator

  School of Computing, Newcastle University, 2021-2025

  Supported MSc Deep Learning students with debugging, experiment design, and evaluation.

  Improved student proficiency in Neural Networks and Computer Vision experimentation.

💻 Skills

• Deep Learning & Machine Learning: PyTorch, TensorFlow, Keras, Scikit-learn, Convolutional Neural Networks (CNNs), Transfer Learning, Fine-Grained Classification, Object Detection, Segmentation, Generative AI, Latent Diffusion Models (LDMs), ControlNet, Vision-Language Models (VLM), Model Robustness, Domain Adaptation, Explainable AI (XAI), Data Augmentation Frameworks, Synthetic-to-Real Generalisation
• Computer Vision: YOLOv5/YOLOv8, OpenCV, Image Processing Pipelines, Dataset Generation and Annotation, Evaluation Framework Design, Domain Shift Analysis, Image Quality Assessment, Image-to-Image Translation (Pix2Pix)
• Programming & Tooling: Python, Java, SQL, Git, Linux, Docker, Bash Scripting, Jupyter, Experiment Tracking, Gradio, Hugging Face Ecosystem
• Synthetic Data & Simulation: Unreal Engine 5 (Blueprints \& C++), AirSim, Automated Dataset Generation, Weather Simulation, Scripted Environments
• Data Analysis: NumPy, Pandas, Matplotlib, Seaborn, Web Scraping, Unstructured Data Processing
• Soft Skills: Technical Writing, Team Collaboration, Public Speaking, Problem Solving

📚 Publications

• Mobile phone image-based framework for anti-copy pattern detection and classification

  Authors: Joseph Smith, Zheming Zuo, Jonathan Stonehouse, Boguslaw Obara

  Journal: Array

• An Augmentation-based Model Re-adaptation Framework for Robust Image Segmentation

  Authors: Zheming Zuo, Joseph Smith, Jonathan Stonehouse, Boguslaw Obara

  Conference: European Conference on Computer Vision

  

• Robust and Explainable Fine-Grained Visual Classification with Transfer Learning: A Dual-Carriageway Framework

  Authors: Zheming Zuo, Joseph Smith, Jonathan Stonehouse, Boguslaw Obara

  Conference: IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2024

  Pages: 183--193

  

  View Paper on arXiv

• How Quality Affects Deep Neural Networks in Fine-Grained Image Classification

  Authors: Joseph Smith, Zheming Zuo, Jonathan Stonehouse, Boguslaw Obara

  Conference: International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, 2024

  Pages: 448--457

  

  View Paper on arXiv

🎤 Presentations

• Robust and Explainable Fine-Grained Visual Classification with Transfer Learning: A Dual-Carriageway Framework

  Conference: CVPR Workshop on Fair, Data-Efficient, and Trusted Computer Vision

  Location: Seattle Convention Centre, Seattle, USA

  Date: Summer 2024

• How Quality Affects Deep Neural Networks in Fine-Grained Image Classification

  Conference: VISAPP

  Location: Precise House Mantegna Roma, Rome, Italy

  Date: Winter 2024

📞 Contact