Computer Vision Lab
Work with usAbout us
We are the Computer Vision Lab of the Faculty of Engineering and Applied Sciences at Universidad de los Andes, Chile. We study self-supervised and multimodal learning so that vision models need fewer labels, generalize better and can explain themselves, and we apply them where they matter: medical imaging, sketch-based understanding, historical documents and video. We work with hospitals, companies and universities in Chile and abroad.
Meet the team →Upcoming event
V Workshop on Computer Vision 2026
AI at the service of the patient: advances and challenges
- When
- Wednesday 9 December 2026, 09:00–17:00
- Where
- Auditorium B, Library, Universidad de los Andes, Chile

News
-

New demo: Search Motor
Search an e-commerce catalogue by drawing a product or showing a photo of it — skDINO for sketches, a fine-tuned CLIP for photos.
-
Help us build a sketch dataset
Help us build a sketch dataset
Sketch Gatherer shows you an object and asks you to draw it; every stroke is saved with its timing for our sketch models.
-

skDINO published in CVIU
“DINOv3 meets sketches: a general sketch-photo encoder”, in Computer Vision and Image Understanding. Code, results and figures on the project page.
-

Two papers at MICCAI 2026 workshops
Real-time low-dose CT denoising with single-step flow matching (MIRASOL) and biased masked image modelling for fine-grained anatomy segmentation (MedAGI).
-
New Fondecyt Regular project
New Fondecyt Regular project
Leveraging visual latent spaces and large language models to make vision models interpretable and explainable.
-

IV Workshop on Computer Vision
Keynotes by Yi-Zhe Song (Surrey), Gilberto Ochoa (Tec de Monterrey), Laurent Heutte and Simon Bernard (Rouen), plus student talks and posters.
Research lines
All research →-

Medical Imaging
AI is advancing fast in language and vision, and healthcare stands to benefit the most. We build medical imaging models that are multimodal,...
Read more → -

Sketch-based Understanding
Sketching is one of the most natural ways people communicate. We build encoders that understand sketches and connect them with photos: for retrieval,...
Read more → -

Efficient One-Shot Detection
Detecting objects from classes the model has never seen, given a single example — even when that example is a sketch.
Read more → -

Generative Models
Conditional generative models — from GANs to diffusion — that turn a sketch into a realistic image, or a low-dose CT into a diagnostic-quality one.
Read more → -

Video Analysis
Models that describe video in the wild — news, sports, streaming — and that let you search video catalogues with a sketch.
Read more →
Latest publications
All publications →-
DINOv3 meets sketches: A general sketch-photo encoder
Computer Vision and Image Understanding (CVIU), 2026
-
LDCT-to-SDCT as a Bridge Problem: Single-Step Residual Endpoint Flow Matching for Real-Time Denoising
To appear in MIRASOL, MICCAI, 2026
-
B-MIM: Biased Masked Image Modeling for Generalizable Segmentation of Fine-Grained Anatomical Structures
To appear in MedAGI, MICCAI 2026
-
A Study of Cross-Dataset Generalization on Liver Vessel Segmentation Improved by a Topological Loss
Medical and Biological Engineering and Computing (2026)