Obz AI Logo
Show your love for obzai with a star!

The open standard for computer vision monitoring

Monitor, explain, and improve your AI vision systems with comprehensive logging and powerful visualization tools. Built for data scientists and ML engineers.

$ pip install obzai
View on GitHubPyPI Package
Open Source
Active Development
Growing Community

obzai package features:

A comprehensive toolkit for monitoring, explaining, and improving computer vision systems with cutting-edge AI observability features.

Explainable AI (XAI)

Generate visual explanations for your computer vision models with state-of-the-art XAI techniques. Understand what your models see and why they make specific decisions.

Model Monitoring

Track model performance, data drift, and system health in real-time. Get alerted to performance degradation before it impacts your users.

Outlier Detection

Automatically identify anomalous inputs, edge cases, and data quality issues that could affect your model performance.

Vision-First Design

Built specifically for computer vision workflows including CNNs, Vision Transformers (ViT), and other deep learning architectures.

Offline Capability

Run entirely offline for secure environments. No data leaves your infrastructure while still getting powerful insights.

PyTorch Integration

Seamlessly integrate with your existing PyTorch workflows. Works with your current training and inference pipelines.

Interactive Dashboard

Connect to a web dashboard with just a few lines of code for quick visualization and analytics of your model behavior.

Data Profiling

Generate comprehensive profiles of your image datasets, tracking distributions, quality metrics, and potential biases.

AI Safety

Ensure your vision systems are fair, unbiased, and compliant with regulatory requirements through comprehensive analysis.

Ready to make your vision AI systems more transparent and reliable?

$ pip install obzai

Getting Started

Get the obzai package up and running in your vision AI pipeline in minutes. Choose your integration path and start monitoring immediately.

Get started in seconds with pip

# Install obzai pip install obzai # Or install from GitHub for latest features pip install "git+https://github.com/obzai/obzai"

Need help getting started? Check out our comprehensive documentation.

📚 Documentation🚀 Quickstart Guide🔧 GitHub Repository📦 PyPI Package

What can you do with the obzai package?

From explainable AI to production monitoring, the obzai package provides comprehensive tools for every stage of your computer vision pipeline.

XAI Tools

Generate visual explanations for your computer vision models using attention maps, CDAM, and saliency techniques.

  • Attention maps for ViT models
  • CDAM gradient-based explanations
  • SaliencyMap with SmoothGrad
  • Region-based XAI visualizations

Data Inspector

Monitor input data quality, detect outliers, and ensure consistent model performance with advanced feature extraction.

  • First-order feature extraction
  • GMM-based outlier detection
  • Deep learning embeddings with PCA
  • Data drift monitoring

Dashboard Integration

Connect to the Obz AI cloud platform for comprehensive monitoring and collaborative workflows.

  • ObzClient for seamless logging
  • Project-based organization
  • Real-time monitoring dashboard
  • Cross-team collaboration

XAI Evaluation

Assess the quality and reliability of your explainability methods with fidelity and compactness metrics.

  • Fidelity scoring for accuracy
  • Compactness evaluation
  • Perturbation-based testing
  • Method comparison tools

Core Capabilities

Comprehensive tools for explainable AI, data quality monitoring, and production deployment

Visual Explanations

Generate and visualize attention maps, saliency maps, and gradient-based explanations

Outlier Detection

Identify anomalous inputs and out-of-distribution samples automatically

Cloud Dashboard

Monitor models in production with the Obz AI platform and collaborative tools

Evaluation Metrics

Assess XAI quality with fidelity and compactness scoring systems

XAI Tools Comparison

AttentionMap

ViT Models
CNN Models
Gradient-Based

CDAM

ViT Models
CNN Models
Gradient-Based

SaliencyMap

ViT Models
CNN Models
Gradient-Based

XAI Regions

ViT Models
CNN Models
Gradient-Based

All XAI tools support region-based analysis and can be evaluated with fidelity and compactness metrics.

Join the obzai Community

Be part of a growing community that comes together to make AI technology more transparent, reliable, and accessible for computer vision.

GitHub Repository

View the source code, contribute to the project, and report issues on our main repository.

View Repository

Complete Documentation

Comprehensive guides covering installation, XAI tools, data inspection, and dashboard integration.

Read Docs

Quickstart Guide

Get up and running quickly with step-by-step instructions for common computer vision workflows.

Get Started

Contact Support

Need help with implementation or have questions? Reach out to our team for assistance.

Ready to Get Started?

Join the growing community of developers building transparent and reliable computer vision systems with the obzai package.

View on GitHubTry the Quickstart
“AI engineers need better tools to ensure transparent and reliable computer vision systems. The obzai package makes it easy for developers to monitor, explain, and improve their vision models.”
The Obz AI Team
Building the future of explainable AI

Contributing to obzai

We welcome feedback and contributions from the community! Whether you're reporting issues, sharing feedback, or helping other users, your input helps make the obzai package better for everyone.

Report Issues

Found a bug or have a feature request? Help us improve the obzai package by reporting issues.

  • 1
    Search existing issues to avoid duplicates
  • 2
    Create a detailed bug report with reproduction steps
  • 3
    Include your environment details (Python version, OS, etc.)
  • 4
    Provide sample code or data when possible

Share Feedback

Help us understand how you use the obzai package and what improvements would be most valuable.

  • 1
    Share your use cases and workflows
  • 2
    Suggest new features or improvements
  • 3
    Report usability issues or pain points
  • 4
    Participate in community discussions

Improve Documentation

Help make the obzai package more accessible by improving docs, examples, and tutorials.

  • 1
    Fix typos and unclear explanations
  • 2
    Add more comprehensive examples
  • 3
    Create tutorials for specific use cases
  • 4
    Improve API documentation

Help the Community

Help other users learn and succeed with the obzai package in their computer vision projects.

  • 1
    Answer questions in GitHub discussions
  • 2
    Share your implementation examples
  • 3
    Mentor new contributors
  • 4
    Create and share best practices

Community Guidelines

Follow these guidelines to ensure effective collaboration and maintain our community standards

Professional Communication

We maintain respectful and constructive communication in all interactions.

Detailed Reporting

Provide comprehensive information when reporting issues or suggesting features.

Quality Focus

Help us maintain high standards for documentation and user experience.

Collaborative Approach

Work together to build the best possible computer vision monitoring tools.

Ready to Get Involved?

Start by exploring our documentation and reach out to our team with any questions or feedback.

Report IssuesDocumentation

Questions or Need Help?

The Obz AI team is here to help you succeed with the obzai package. Don't hesitate to reach out with questions or feedback.

🐛 Report Issues🚀 Quick Start Guide📚 Full Documentation