GLIMS-MedNeXt
Brain tumor segmentation under domain shift
PhD Student and Researcher · Computer Vision
Ali
Azmoudeh.
I build and evaluate computer vision and multimodal models—from brain MRI segmentation to synthetic facial data and vision–language understanding.
Istanbul Technical University · SiMiT LabScroll to discover
One question. Three connected directions.
From seeing.
To understanding.
How can a model generalize responsibly when the real world does not match its training set?
MULTIMODAL VISION / 02VISUAL + LANGUAGE
FACEvisual analysis
BRAINmedical imaging
VISION–LANGUAGEmultimodal reasoning
Research imagery is conceptual, not experimental output.
Medical image analysis
Reliable vision.
When it matters.
Brain MRI segmentation across high- and low-quality imaging domains.
Beyond clean benchmarks
Different data.
The same responsibility.
I explore transfer learning, ensemble models, and conformal risk control to study reliability under domain shift.
Facial analysis & synthetic data
Expressions.
New perspectives.
Understanding expressions through real and synthetic visual data.
Learning from generated data
More possibilities.
Thoughtful evaluation.
Privacy-aware facial expression recognition and text-guided expression generation using vision-language and generative models, synthetic data, and cross-dataset evaluation.
Multimodal visual understanding
Images meet
language.
Connecting what a model sees with what language means.
Visual + language
Different modalities.
Shared understanding.
Vision-language models and visual-textual cross-attention for multimodal reasoning, including IMMCAN idiom understanding and open-world text-guided visual counting.
Technical Expertise
From research questions
to working models.
- Python
- PyTorch
- MONAI
- OpenCV
- Hugging Face Transformers
- Vision-Language Models
- Generative AI
- CUDA
- Git
- Linux
- HPC
Computer vision & model development
Build training and evaluation pipelines for image segmentation, facial expression recognition, and multimodal representation learning. Adapt pretrained models, compare architectures, and evaluate generalization across datasets.
Experimentation & compute
Run GPU-based research workflows in Python and PyTorch, with medical-image processing in MONAI and experiments on Linux and HPC systems. Analyze accuracy, segmentation quality, uncertainty, and failure cases.
Language + vision
Multimodal AI,
grounded in experiments.
My applied work connects language representations with visual tasks: image ranking, idiom understanding, and text-guided facial expression generation.
Explore the IMMCAN case study- Pretrained language and vision–language encoders
- Cross-attention and multimodal feature fusion
- Generative models for synthetic visual data
- Hugging Face Transformers and prompt engineering
Additional areas of knowledge include large language models and retrieval-augmented generation (RAG), alongside graduate study in NLP, advanced NLP, and generative AI.
Selected project case studies
The work behind
the research.
Read case study
SyntFER
Learning facial expressions from synthetic data
IMMCAN
Connecting idioms, captions, and images
Selected publications
Ideas into
contributions.
Peer-reviewed work and open research across medical vision, facial analysis, multimodal reasoning, and visual counting.
P / 01
On Applicability of Synthetic Datasets for Facial Expression Recognition
2026 IEEE 20th International Conference on Automatic Face and Gesture Recognition (FG) · 2026
Kyoto, Japan · pp. 1–10 · First author
Evaluates pseudo-labeling, diffusion synthesis, and GAN expression editing for facial expression recognition, including how synthetic and real data combine under cross-dataset evaluation.
P / 02
VisAffect at MWE-2026 AdMIRe 2: IMMCAN Idiom Multimodal Cross-Attention Network
Proceedings of the 22nd Workshop on Multiword Expressions (MWE 2026) · 2026
Association for Computational Linguistics · pp. 149–153 · Equal contribution with Barış Bilen
Studies multilingual image ranking using frozen language and vision–language encoders with a trainable cross-attention module. Visual grounding improves zero-shot transfer, while text-only modeling performs better on the smaller in-domain test.
P / 03
GLIMS-MedNeXt: An Ensemble Framework for Brain MRI Segmentation in Sub-Saharan Africa
Segmentation, Classification, and Synthesis for Brain Tumors and Traumatic Brain Injuries · 2026
MICCAI 2025 challenges: BraTS-Lighthouse · LNCS 16376 · Springer, 2026 · pp. 262–273 · First author
Combines GLIMS and MedNeXt with pretraining, fine-tuning, and ensemble fusion for brain-tumor segmentation in lower-quality MRI.
P / 04
A Survey on Class-Agnostic Counting: Advancements from Reference-Based to Open-World Text-Guided Approaches
Computer Vision and Image Understanding · 2026
Volume 267 · Article 104703
Organizes 29 counting approaches into reference-based, reference-less, and open-world text-guided methods, comparing performance and generalization on FSC-147 and CARPK.
P / 05
Advanced Facial Expression Classification with CNN-Transformer Integration for Human-Computer Interaction
2024 9th International Conference on Computer Science and Engineering (UBMK) · 2024
Antalya, Türkiye · pp. 800–805 · First author
Applies facial detection, landmark and feature extraction, and CNN–Transformer classification to examine expression responses during online shopping tasks.
Behind the research
Curiosity.
Learning.
Problem solving.
I am a PhD student in Computer Engineering at Istanbul Technical University and a Graduate Research Assistant at SiMiT Lab, working under the supervision of Prof. Dr. Hazım Kemal Ekenel.
My research sits at the intersection of computer vision, deep learning, and trustworthy AI. I study how vision models behave beyond clean benchmarks: under domain shift, limited annotations, low-quality medical scans, privacy constraints, and imbalanced data.
Current directions include reliable brain-tumor MRI segmentation, conformal risk control, vision-language-guided facial expression generation, multimodal idiom understanding, and class-agnostic visual counting.
I am enthusiastic about exploring emerging ideas, continuously expanding my knowledge, and applying what I learn to solve challenging real-world problems.
- Based
- Istanbul, Türkiye
- Affiliation
- ITU · SiMiT Lab
- Role
- PhD Student and Researcher · RA/TA
- Advisor
- Prof. Dr. Hazım Kemal Ekenel
- Focus
- Computer Vision · Multimodal AI · Trustworthy Medical AI
Research & education
The path so far.
2025—
PhD, Computer Engineering
Istanbul Technical University · Computer vision and trustworthy medical AI
2024—
Graduate Research & Teaching Assistant
Istanbul Technical University · Supported by the ITU–Turkcell research scholarship
2022—
Graduate Research Assistant
SiMiT Lab, ITU · Computer vision, biometrics, and medical imaging
2022–25
MSc, Computer Engineering
Istanbul Technical University
2016–21
BSc, Computer Engineering
University of Tabriz
Research collaboration
Let’s build vision systems that generalize.
For research discussions, collaborations, or questions about my published work, reach me through my ITU email.
Start a conversation