Software
Idiap has a successful track record of dissemination of data and software that are widely adopted
This commitment extends beyond mere software contributions. Idiap actively participates in numerous projects, ranging from small-scale initiatives to high-profile endeavors. To facilitate the sharing of improvements made by Idiap's researchers with the wider community, the institute has entered into various corporate contributor license agreements (CLA) or equivalent arrangements with The Qt Company, The Python Project, The Cloud Native Computing Foundation, The Cloud Foundry Foundation, and Google. Thanks to these CLAs, Idiap employees are able to contribute to large-scale open-source projects.
Idiap creates and makes available a significant number of professional software. Over the past five years, Idiap has filed 206 software disclosures, which enabled the distribution of 122 open source software packages and granted 84 commercial licenses on patents and software.
Examples of software packages created by Idiap are Fast Transformers, PyDHN, Kaldi and Bob. The Github project Fast Transformers, which has over 1400 stars and over 160 forks, was used in Muzic, a Microsoft project for music understanding and generation. The open-source package PyDHN for the physics-based simulation of district heating networks, computes pressure, temperature, and mass flow within the pipework knowing only the boundary conditions at the central(s) and sub-stations; it is included in open-source GIS tools for a wider impact on society. The Kaldi open source tool for speech modeling, which is considered one of the main technologies in the community for research and innovation, received 6781 citations. Bob, a set of open-source tools to promote reproducible research, has 25 releases, over 100 satellite repositories and over 5000 commits for the core repository.
List of softwares
Name | Description | Date |
|---|---|---|
Evaluating Multimodal Large Language Models for Heterogeneous Face Recognition | A systematic evaluation of state-of-the-art MLLMs for heterogeneous face recognition (HFR), where enrollment and probe images are from different sensing modalities, including visual (VIS), near infrared (NIR), short-wave infrared (SWIR), and thermal camera. | |
Synthetic Data Generation for Low-Resolution Face Recognition | Code for "Improving Low-Resolution Face Recognition under Limited Data:
How Synthetic Data Generation Can Close the Domain Gap" at IJCB 2026
Focus Session "Generative AI for Fair and Secure Biometrics under
Limited Data". | |
prism | Prism is a typed programming language that makes reasoning,
capabilities, and assurance boundaries explicit in systems built with
generative AI. | |
mech-reasoner | Mech Reasoner is a benchmark toolkit for qualitative mechanistic reasoning. It validates electrical, mechanical, and thermal mechanism catalogs; generates open-answer JSON tasks; evaluates structured model answers; and plots accuracy by task complexity. | |
DriveFace | DriveFace: A Cross-Spectral Through-Glass Face Dataset for On-the-Move Vehicular Border Control | |
NVDP-Clipping | Nonparametric Variational Differential Privacy via Embedding Parameter Clipping | |
quantifying-membership-inference | Quantifying Training Membership Information in the Hyperspherical Embedding Geometry of Face Recognition Models | |
LARM | Loop Audio Recurrent Model | |
miccai-omia-beyond-the-last-frame | Beyond the Last Frame: Temporal Modelling of Fluorescein Angiography for Hyperfluorescence Classification | |
PrivLEX | Detecting legal concepts in images through Vision-Language Models |
