# advboxes/AdvBox

Advbox is a toolbox to generate adversarial examples that fool neural networks in PaddlePaddle、PyTorch、Caffe2、MxNet、Keras、TensorFlow and Advbox can benchmark the robustness of machine learning models.  Advbox give a command line tool to generate adversarial examples with Zero-Coding.

Repository: https://github.com/advboxes/AdvBox
Canonical: https://ross.abutalabs.com/products/advbox
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Topics: adversarial-examples, paddlepaddle, machine-learning, security, deep-learning, adversarial-example, onnx, graphpipe, fgsm, adversarial-attacks, deepfool
Last push: 2023-02-15T19:57:27+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2947, "days_push": 1295, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1404, forks 266 (observed 2026-08-28T04:04:37.998440+00:00)

## What it is
AdvBox is a Baidu open-source toolbox for generating adversarial examples that fool neural networks across frameworks like PaddlePaddle, PyTorch, TensorFlow, Keras, MxNet, and Caffe2. It also benchmarks model robustness and includes tools for detecting adversarial examples and studying data poisoning, with a zero-coding CLI for generating attacks.

## Use cases
- generate adversarial examples to fool image classifiers
- benchmark the robustness of machine learning models
- test whether a cloud image classification service is vulnerable to attacks
- detect adversarial examples in large datasets
- study data poisoning attacks on ML models
- attack face recognition systems in research settings
- generate adversarial examples without writing code via CLI

## When to choose
- you need a multi-framework adversarial attack toolbox with a zero-coding CLI
- you are doing AI security research on attack and defense of neural networks
- you want to benchmark model robustness against FGSM, DeepFool, and similar attacks
- you work in PaddlePaddle and need a lightweight adversarial SDK

## When to avoid
- you need actively maintained tooling with recent updates and community support
- you want adversarial training defenses integrated into a modern training pipeline
- you need support for the latest PyTorch or TensorFlow versions out of the box

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, security, cli, benchmarking
- domain: machine-learning, security, deep-learning, artificial-intelligence
- platform: python, cross-platform
- tags: adversarial-examples, adversarial-attacks, adversarial-defense, model-robustness, paddlepaddle, pytorch, tensorflow, fgsm, deepfool, data-poisoning

## Member repositories
- advboxes/AdvBox (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:37.998440+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:38:49.746044+00:00, confidence not recorded.
  - readme: https://github.com/advboxes/AdvBox (fetched 2026-08-28T04:04:37.998440+00:00, sha 9eaaa3eaa9c7)
- Data as of 2026-08-30T08:39:29.467469+00:00.