My Portfolio
Custom Climbing Grip Trainer
Personal project — Computer Vision
End-to-end pipeline that measures hand proportions from a single photo and generates a custom-fit ergonomic grip trainer.
- RF-DETR and SAM2 for object detection and segmentation
- Custom heatmap-based keypoint detector to locate finger creases
- Parametric CAD model generated to the measured hand geometry
- Self-validation checks and interactive feedback guide users toward a good photo and flag bad inputs.
Local Cluster Cardinality Estimation for Adaptive Mean Shift
PhD work
Research article on the adaptive mean shift algorithm based on distance distribution analysis from PhD studies
- Developement of a local cluster cardinality estimator based on distance distributions
- Adaptive mean shift with local cardinality estimator
- Various mathematical proofs on distance distributions
Keypoint Masking for Analyzing Segmented Medical Image Data
Master's Thesis
Analysis of keypoint extraction on masked images resulting in an extraction procedure limiting masking related noise.
- Supported by a theorical model valid for images of any dimensions
- Python code
- 3D Medical Images
Climbing Grade Analysis
Vulgarized article
Analysis of difficulty range within Fontainebleau climbing grades: some grades contain twice the difficulty range leading to an uneven grade progression for climbers.
- Database manipulation with Pandas
- Original data analysis leading to new insight
About Me
I completed a Master's degree with a thesis in computer vision. My passion is research involving computer vision. I love gaining a good understanding of a research problem through data analysis and mathematics.
My resume (english)→ Mon CV (français)→
