You want to discover some of our cutting edge technologies?
- Option 1: look at the list below
- Option 2: use the search bar on top of the page
- Option 3: browse by area using the left side menu
- Option 4: find by tag
List of our technologies
We introduce a figure of merit that is able to evaluate and compare fairness aspects between multiple biometric verification systems, the so-called Fairness Discrepancy Rate (FDR). A use case with two synthetic biometric systems is introduced and demonstrates the potential of this figure of merit in extreme cases of demographic differentials.
CBI-MMTools is a set of plugins, device adapters and libraries for the operation of microscopy platforms using Micro-Manager.
Algorithm that automates the segmentation of eye fundus (RGB) images.
Algorithm that automates the construction of Hypnograms from Polysomnograms.
We explore the use of short wave infrared (SWIR) imaging for Face Presentation Attack Detection (PAD). Face PAD is performed using recent models based on Convolutional Neural Networks using only carefully selected SWIR image differences as input.
We propose EdgeFace, a lightweight and efficient face recognition network inspired by the hybrid architecture of EdgeNeXt.
We propose SynthDistill, a new framework to train lightweight face recognition models by distilling the knowledge of a pretrained teacher face recognition model using synthetic data.
Toolbox for efficient learning and control in robotics by relying on geometric algebra
The method is a novel differentially private Graph Neural Network (GNN) based on Aggregation Perturbation (GAP), which can statistically obfuscate the presence of a single edge (edge-level privacy) or a single node and all its adjacent edges (node-level privacy), thus providing formal privacy guarantees with competitive classification performance.
We introduce a novel Conditional Adaptive Instance Modulation (CAIM) module that can be integrated into pre-trained FR networks, transforming them into HFR networks.
We propose a surprisingly simple, yet, very effective method for matching face images across different sensing modalities.
The Idiap Human Perception system is a real-time multimodal system for person tracking, face re-identification, sound source localisation, and visual focus of attention.
CAD that uses Chest X-ray Imaging for detecting Active Pulmonary cases.
The method combines Local Differential Privacy and Graph Neural Networks (GNNs) to preserve the privacy of node features and labels in GNNs, with relatively low privacy cost while providing competitive accuracy, using differential privacy principles.
PolyProtect transforms a face embedding to a more secure template, using a mapping based on multivariate polynomials parameterised by user-specific coefficients and exponents.
Toolbox to solve optimal control and trajectory optimization problems in robotics.
Our method allows estimating the parameters of a spatially variant Point Spread Function (PSF) model using a Convolutional Neural Network (CNN).
We have pioneered the first viable deep learning framework (task definition, network architecture, training paradigm) for solving fundamental auditory tasks such as sound source localization, speaker identification and speech/non-speech classification.
Software enabling hybrid artificial and spiking networks.
We introduce a method to generate a synthetic dataset, without the need for human intervention, by exploiting the latent structure of a StyleGAN2 model with multiple controlled factors of variation.
Toolbox to solve global optimization problems in a data-efficient manner
State of the art neural synthesisers demonstrating Idiap's expertise in speech synthesis: localisation of accents and dialects, synthesis of emphasis and emotions.
We used transfer learning from the vision transformer model for the zero-shot presentation attack detection task.
Automatic assessment of sign language production with clues on how to improve.