From raw spectra to deployed models — a comprehensive Python framework and desktop application for Near-Infrared Spectroscopy, built by researchers, shared with the world.
Most people meet nirs4all in one of two ways: the Python library that powers everything, or Studio, the no-code app built on top of it. Everything else — file readers, datasets, portable engines, a cross-language port — is the supporting ecosystem, mapped out further down.
A comprehensive Python framework for Near-Infrared Spectroscopy analysis. Build declarative ML pipelines with 30+ preprocessing transforms, 15+ advanced PLS models, and multi-backend support for scikit-learn, TensorFlow, PyTorch, and JAX. Export trained pipelines as portable .n4a bundles ready for deployment.
A modern desktop application for interactive spectral data exploration and model building. Features a drag-and-drop pipeline builder, real-time spectral visualization, SHAP variable importance analysis, and a native desktop experience powered by Electron — with GPU acceleration support on Windows, Linux, and macOS.
Interactive desktop application for exploring data, building pipelines, and analyzing results — no code required.
The Playground — interactive spectral visualization with real-time preprocessing, distribution charts, stacked composition views, and PCA analysis.
Model ranking with cross-validation scores, RMSE, R², and performance metrics — compare dozens of pipeline variants side by side.
Runs overview — track all experiments, filter by status, and navigate your workspace history.
Predictions page — visualize predicted vs. actual values, inspect residuals, and export results.
The Inspector — deep-dive into individual predictions, spectral contributions, and outlier detection across samples.
The Pipeline — configure and manage your data processing pipelines with ease.
The SHAP — visualize feature importance and contributions for model predictions.
The Synthesis — allow users to generate synthetic data and explore hypothetical scenarios.
Built specifically for NIRS workflows — not generic ML tools adapted for spectra.
Comprehensive NIRS-specific preprocessing including wavelength-aware operators that automatically receive wavelength information.
State-of-the-art Partial Least Squares variants with automatic operator selection via AOM-PLS for optimal model performance.
Unified API across all major ML frameworks. Swap backends without changing your pipeline code.
Express complex workflows — branching, merging, hyperparameter sweeps, cross-validation — with a concise, readable list syntax. Parallel execution via joblib.
Full scikit-learn compatibility enables SHAP integration for spectral band importance visualization, prediction diagnostics, and model transparency.
Export entire trained pipelines as .n4a bundles for seamless deployment and sharing. Includes predict, explain, and retrain capabilities.
Near-infrared spectroscopy provides rapid, non-destructive analysis across industries — from field to laboratory.
Crop quality assessment, soil composition analysis, and phenotyping for plant breeding programs.
Rapid authentication, composition analysis, and quality grading of food products and raw materials.
Pharmaceutical process monitoring, chemical analysis, and materials characterization in research laboratories.
Install the Python library from PyPI — no configuration required. Add optional extras for deep-learning backends. (For the no-code route, grab Studio instead.)
# Core library pip install nirs4all # With PyTorch backend pip install nirs4all[torch] # With TensorFlow backend pip install nirs4all[tensorflow] # All backends (sklearn + TF + PyTorch + JAX) pip install nirs4all[all] # GPU/CUDA acceleration pip install nirs4all[all-gpu]
import nirs4all # Run a pipeline on your spectral data result = nirs4all.run( pipeline=[SNV(), PLSRegression(10)], dataset="data.csv" ) # Inspect results print(result.best_rmse) print(result.top(5)) # Deploy the best model result.export("best_model.n4a") # Predict on new spectra preds = nirs4all.predict("best_model.n4a", new_data)
import nirs4all # Preprocessing variants × PLS sweep on n_components pipeline_1 = [ {"_or_": [SNV, MSC, Detrend]}, {"model": { "type": PLSRegression, "param": {"n_components": {"_range_": [2, 20, 5]} } } ] # Stacking: compare PLS vs RF in branches, then merge + meta-learner pipeline_2 = [ {"branch": [ [SNV(), {"model": PLSRegression(10)}], [MSC(), {"model": RandomForestRegressor()}], ]}, {"merge": "predictions"}, {"model": Ridge()}, # meta-learner trained on stacked predictions ] results = nirs4all.run(pipeline=[pipeline_1, pipeline_2], dataset="data.csv")
At the center sits nirs4all, the Python library. Around it: the no-code Studio app, in-browser demos, file & dataset tools, portable engines, a leakage-safe execution core — and an ongoing port to every language. Explore the live demos, or install any package below.
The Python core of the ecosystem: declarative NIRS pipelines with 30+ spectral preprocessings, 15+ advanced PLS models, and a unified API over scikit-learn, TensorFlow, PyTorch and JAX. Train, explain, and export portable .n4a model bundles.
The no-code application built on the library: a drag-and-drop pipeline builder, real-time spectral visualization, experiment tracking, and SHAP analysis — packaged with Electron for Windows, Linux and macOS, and runnable on the web.
A portable Partial Least Squares engine in C++17 with a stable C ABI (libn4m) and first-class bindings for Python, R, MATLAB/Octave, JavaScript/WebAssembly and Android — one numerical core, parity-checked across every language.
Operator-adaptive calibration models — AOM-PLS, POP-PLS and AOM-Ridge — that fold spectral preprocessing directly into the model and replace external preprocessing grid-search. scikit-learn compatible; companion code for the AOM-PLS paper.
Rust-first, low-level readers for ~58 NIRS & spectroscopy format families. Content-sniffed, lossless, provenance-tracked records with Python, R, WebAssembly and C bindings — the messy vendor-file zoo turned into one clean data model. Try the full reader catalog right in the browser via the live demo.
A dataset-assembly bridge: turn any input — a folder, a glob, vendor spectra plus a reference table, or a config — into a pipeline-ready dataset through resolve → infer → configure → materialize, with a score-based inference engine. Built on nirs4all-formats.
A Python library that transparently pulls curated, hand-picked NIRS test datasets from Dataverse — turning reproducible, DOI-citable benchmarks and lab experiments into a one-line import.
A leakage-safe, in-process DAG execution core. It owns the graph, phases, folds, out-of-fold joins, lineage, caching and deterministic RNG — exposed through a C ABI so any host language can drive reproducible ML pipelines.
The data-contract and planning layer beneath dag-ml: typed, sample-aligned, multi-source data views, representation adapters, data plans, and schema fingerprints — the foundation that keeps pipelines reproducible.
A public alpha prototype for distributing nirs4all.run() across lab machines — FastAPI coordinator, SQLite queue, capability routing and crash-safe leases. It is a validation bench, not a production service.
The browser-native client: a full-WASM mini Studio that runs the whole NIRS loop — load spectra, build a pipeline from nirs4all-methods nodes, train, score and predict — entirely client-side, no Python.
The canonical low-level distribution that aggregates dag-ml, dag-ml-data, formats, io, datasets and methods. It exposes native binding surfaces without adding parsers or methods; full upstream integrations and pipeline parity fixtures are the next hardening step.
Reproducible, scored nirs4all pipelines run on curated reference datasets, published as a browsable benchmark resource. The design is taking shape; it is not a public submission platform.
Researchers and engineers from the PHENOMEN team at CIRAD / UMR AGAP Institut, building open-source tools for spectroscopy science.
nirs4all is developed at CIRAD within the UMR AGAP Institut — a research center dedicated to plant genetics, genomics, and agro-resources.
The French Agricultural Research Centre for International Development — a research organization working with developing countries to tackle international agricultural and development challenges.
Visit CIRAD →
Amélioration Génétique et Adaptation des Plantes méditerranéennes et tropicales — a joint research unit focused on plant genetic improvement, genomics, and adaptation for Mediterranean and tropical species.
Visit UMR AGAP →Research institutes, universities, and companies across Europe and Africa build their near-infrared spectroscopy workflows with nirs4all.
Research using nirs4all and related spectroscopic analysis.
@software{beurier2025nirs4all, author = {Gregory Beurier and Denis Cornet and Lauriane Rouan}, title = {NIRS4ALL: Open spectroscopy for everyone}, url = {https://github.com/GBeurier/nirs4all}, version = {0.8.8}, year = {2026}, }
Answers to the main questions researchers, engineers, and applied teams ask before adopting nirs4all for Near-Infrared Spectroscopy workflows.
nirs4all is open-source Near-Infrared Spectroscopy software made of a Python library and a desktop application for preprocessing spectra, building chemometric and machine-learning pipelines, visualizing results, and deploying models.
Yes. nirs4all Studio provides a desktop interface for interactive exploration, pipeline building, experiment tracking, result analysis, and data inspection without needing to write code.
The nirs4all Python library supports multiple backends and integrations including scikit-learn, TensorFlow, PyTorch, JAX, and advanced chemometric workflows around Partial Least Squares models.
You can install the Python library from PyPI with pip install nirs4all, and you can download nirs4all Studio for Windows, macOS, and Linux from the latest GitHub releases page.
Everything you need to get started, contribute, or explore the wider nirs4all ecosystem — from the core library and Studio to the portable engines and file readers underneath.
Source code, issues, and contributions for the Python library
Source code for the nirs4all Studio desktop application
Full API reference, tutorials, and usage guides on Read the Docs
Official Python package index page with release history
The meta-repository pinning every nirs4all and dag-ml project together
Public deposited papers, reproducibility kits, and stable companion artifacts
Source repository for this website and its GitHub Pages deployment
Portable C++/C-ABI PLS engine (libn4m) with bindings for every language
Live demo of the Rust readers for ~58 NIRS & spectroscopy file formats
Live demo of the dataset-assembly bridge that turns any input into a dataset
Catalog site for the curated, DOI-citable NIRS reference datasets
The whole NIRS loop running in your browser, no Python (WASM)
Portable aggregate binding scaffold for Rust, Python, R, MATLAB/Octave and WASM
Operator-adaptive AOM-PLS, POP-PLS & AOM-Ridge calibration models on PyPI
Public alpha prototype for distributed nirs4all execution
Report bugs, request features, or ask questions about the Python library
Report bugs, request features, or ask questions about nirs4all Studio
The French Agricultural Research Centre for International Development
Joint research unit for plant genetic improvement and adaptation
Release notes and version history for the nirs4all library