Aymen Sekhri
Ph.D. candidate in machine learning · University of Poitiers & NTNU
I design trustworthy computer vision systems for immersive media and clinical imaging, bridging perceptual studies with deep learning models.
Research focus
- Machine learning for perceptual image and video quality, with an emphasis on augmented and mixed reality scenarios.
- Deep neural architectures for medical imaging diagnostics, particularly localisation-aware models for musculoskeletal health.
- Human-centric evaluation pipelines that connect subjective experiments with reproducible computational metrics.
Latest publications
Shifting Focus: From Global Semantics to Local Prominent Features in Swin Transformers for Knee Osteoarthritis Severity Assessment
Conventional imaging diagnostics frequently encounter bottlenecks due to manual inspection, which can lead to delays and inconsistencies. Although deep learning offers a pathway to automation and enhanced accuracy, fo...
Do Digital Images Tell the Truth?
Since the advent of digital cameras, image editing tools have made it straightforward to manipulate content. Copy-move forgeries—where a region is duplicated and pasted within the same frame—are particularly challengi...
Automatic Diagnosis of Knee Osteoarthritis Severity Using Swin Transformers
Knee osteoarthritis is a widespread condition that can cause chronic pain and loss of mobility. Early detection and grading are critical for effective intervention, yet manual assessment remains labour intensive.
Complete list of publications →
Research highlights
REALISME – Perceptual Quality for Augmented Reality
Doctoral researcher · 2022 – present
Developing learning-based quality assessment pipelines that connect objective metrics with human perception in augmented reality scenarios.
- Designing subjective and objective protocols that span indoor and outdoor mixed-reality capture.
- Bridging expertise from the University of Poitiers and NTNU to deliver cross-laboratory validation.
Automated Knee Osteoarthritis Assessment
Lead author · 2023 – present
Building transformer architectures that localise clinically salient biomarkers in radiographic images to support early-stage diagnosis.
- Introduced localisation-aware Swin Transformer variants for robust multi-dataset training.
- Collaborating with clinical partners to align predictions with Kellgren–Lawrence grading.
News
- Sep 2024 Presenting our EUSIPCO 2024 paper on local feature refinement for knee osteoarthritis analysis.
- Jan 2024 Published a book chapter on verifying the trustworthiness of digital imagery in Digital Image Security.
- Jul 2023 Shared our Swin Transformer framework for automatic knee osteoarthritis grading at CBMI 2023.
- Oct 2022 Presented our domain adaptation approach for medical image quality assessment at ICIP 2022.
Collaboration
I am co-advised by Mohamed-Chaker Larabi and Seyed Ali Amirshahi and supported by the Nouvelle-Aquitaine Research Council through the REALISME project. I welcome collaborations on perceptual quality assessment, radiology decision support, and applied machine learning for immersive media. Feel free to reach out at aymen.sekhri@univ-poitiers.fr.