Address
48, Boulevard J. Saade
Quai du Lazaret, 13002 Marseille, France
Unlock the full potential of your image, video and audio data
Katomi designs and supports the deployment of computer vision and voice solutions for healthcare, from scientific guidance and project framing to production integration.
Image · Vision & video
Audio · Voice and NLP
MLOps · Deploy
Our offer
A modular offer aligned with your goals
From framing to deployment, we can support one building block or the full cycle: data -> model -> product.
01
Image
Image and video analytics
Extract actionable insight with rigor
- Preprocessing, normalization, enhancement (noise, contrast, artifacts)
- Detection, localization, measurement, counting of objects or structures
- Segmentation of anatomical structures or regions of interest
- Video analysis: clinical events, monitoring, exam quality control
02
Audio
Audio and voice
Unlock the value of healthcare audio
- Automatic transcription and report structuring
- Voice-driven documentation automation
- Context-aware medical speech-to-text
- Clinical entity extraction and form structuring
03
IA
Deep Learning
High-performing models that are evaluated and explainable
- Classification, scoring, estimation in 2D, 3D and time series
- Low-label strategies (semi-supervised, self-supervised)
- Explainability, saliency maps and uncertainty estimation
- Robustness: bias, drift and acquisition variability
04
MLOps
Industrialization and MLOps
From model to product: integration, reproducibility, monitoring
- Reproducible data and training pipelines with versioning and traceability
- Deployment via API, containers, edge or cloud to fit your constraints
- Monitoring, drift detection and controlled retraining
- Full documentation and knowledge transfer
Why Katomi
Beyond tools: 4 reasons to choose us
We can support one component of your challenge or the full development cycle.
01
Healthcare focus
We understand clinical workflows, sensitive data constraints and validation requirements in healthcare.
02
Product co-development
From pilot to industrialization, we work with your R&D, product, clinical and IT teams to deliver a truly integrated solution.
03
Scientific rigor
Performance, limits and conditions of use are documented. We prioritize traceability and explainability.
04
Real-world validation
Multi-site robustness tests, bias control, calibration: a lab-only model is not enough.
Our approach
From framing to industrialization: build, validate, deploy
A pragmatic approach from data strategy to product integration. We iterate in short cycles with clear deliverables at each milestone.
Framing
Feasibility check, data audit, constraints, and definition of the right metrics.
Decide fast
01. Diagnostic
Proof
Rapid prototype, metric-driven iterations, first quantified results on your real data.
Prove value
02. Prototype
Validation
Rigorous evaluation plan, multi-site robustness tests, AI explainability and bias management.
De-risk before prod
03. Validation
Production
Integration into your product or systems, monitoring, incident handling, and continuous improvement.
Deploy and run
04. Production
Okure spotlight
3D modeling of the skin surface.
Capture - Model - Analyze
Okure is Katomi’s internal R&D project. It reconstructs a 3D model of the skin surface from a simple video stream, in real clinical conditions.
A 2D image captures a flat projection. A 3D model reconstructs full geometry: curvature, relief, volume and orientation. This enables precise metric measurements (area, depth, deformation) that are less sensitive to viewpoint and lighting. It supports reliable longitudinal follow-up in a consistent spatial frame.
This project showcases our expertise in 3D reconstruction, real-time video processing and deployment in constrained medical settings.
Project highlights
A simple flow to document and compare skin lesion evolution.
Input
Video stream capture.
Output
3D model generation.
Value
Comparison of lesion evolution over time.
Application domains
Dermatology
Skin wounds and healing
Remote follow-up and decision support.
Technology developed by Katomi - patent protected
Use cases
Concrete projects focused on impact
We adapt to your reality: clinical context, data constraints and product goals.
Lesion detection and segmentation
Medical imaging
Automatic localization of relevant structures, reproducible quantification and case prioritization for radiologists.
Segmentation
3D
Radiologie
Cell classification and counting
Pathology
Histology slide analysis, cell counting, image biomarker extraction for clinical research.
Microscopie
Classification
Biomarqueurs
Visual quality control before submission
Medical devices
Anomaly detection with limited examples, hybrid CV+DL approach, calibrated thresholds and uncertainty.
Inspection
Anomalie
Industrie
Voice-driven form completion
Voice and NLP
Automatic structuring of clinical data from voice, reducing clinicians’ documentation time.
Speech-to-text
NLP médical
Formulaires
Take action
Let’s discuss your use case
Describe your need: we will get back to you with an initial framing, a feasibility view and next steps.
Framing en 45 minutes
Feasibility assessment
Clear action plan
Confidentiality respected
Questions fréquentes
Katomi designs and deploys healthcare image, video and audio solutions from framing to production.
1. Do you need a large volume of data?
Not necessarily. From the start we define a realistic strategy: a conservative baseline, targeted annotation and methods suited to limited labels (semi-supervised, active learning, weak supervision).
2. Can you work with heterogeneous multi-site data?
Yes. Handling heterogeneity is part of our method. We test robustness, address bias during evaluation and document the model’s conditions of use.
3. Do you deliver just a model?
No. Our goal is a deployable solution: integration into your product or systems, reproducible pipelines, performance monitoring over time and documentation. A non-integrable model has little value.
4. Where can development be done?
Depending on your constraints. We adapt architecture and deployment: public cloud, on-premise, isolated environment, edge. Deployment constraints are addressed from the start, not at the end.
5. Who owns the code and deliverables?
Ownership and usage terms are defined contractually from day one. We use clear clauses on IP, usage rights and confidentiality (NDA/DPA).
6. Can you explain the model’s decisions?
When relevant and required, we offer explainability methods (heatmaps, SHAP, Bayesian uncertainty) and document usage limits. We do not promise full transparency where it is not applicable.
Contact
Contact details
Katomi France - reach out by email or via the form.
contact@katomi.co
Tell us about your need
Share the context and goal of your project. We will get back to you quickly.