Computer Vision Consulting

AI & Computer Vision Consulting

Senior, hands-on help with the hard parts of computer vision — object detection, segmentation, co-registration, classification and tracking, designed, trained and shipped to production.

I'm Ruben Cardenes, an independent AI and computer vision consultant based in Munich and working remotely with teams worldwide. Over 20+ years I've built vision systems across medical imaging, digital pathology and defense — from a PhD in medical image analysis to leading AI engineering for digital pathology at scale. If you have a vision problem that's stuck at the prototype stage, or a model that works in a notebook but not in production, that's exactly the kind of work I take on.

What I help teams build

• Object detection & tracking — detecting and following objects in images and video, including small, fast or partially occluded targets (e.g. drone detection, people and vehicle tracking, sports analytics).
• Semantic & instance segmentation — pixel-level understanding for medical, microscopy and industrial imagery, including cell and tissue segmentation.
• Image registration & co-registration — aligning multi-channel, multi-modal or serial images into a common frame.
• Classification & representation learning — from straightforward classifiers to embedding and similarity models.
• Real-time & embedded deployment — optimizing and deploying models to run at speed, including on edge hardware such as NVIDIA Jetson.

Architectures & tools

I choose the model to fit the problem rather than the fashion. Depending on the task that includes transformer-based detectors (DETR, D-FINE), YOLO-family models, U-Net and its variants for segmentation, Vision Transformers, optical-flow methods for motion, and Siamese/metric-learning networks for matching and re-identification.

Where this has been proven

This isn't benchmark-only experience. I've applied computer vision to digital pathology and microscopy (multiplex immunofluorescence, IHC and H&E analysis) while leading AI engineering at Ultivue/Vizgen, to real-time drone detection and video processing at Hensoldt, and to medical imaging research (MRI, CT, DT-MRI) rooted in a decade of academic and postdoctoral work. You can see a selection of open-source work on the projects section of the homepage.