voxel51-eta

What it is and what it does

ETA is an open-source analytics infrastructure for computer vision and machine learning workflows. It provides a core library (eta.core) with utilities for image, video, and embedding manipulation, plus higher-level modules for inference with classifiers and detectors. The package is organized into submodules for different tasks—classifiers, detectors, models, and pipelines—and exposes a command-line interface for building and running analytics workflows without writing code.

The toolkit is designed to be portable across Mac, Linux, and Windows, and supports both CPU and GPU setups. It works with TensorFlow 1.X and 2.X and includes a model registry (manifest.json) that downloads pre-trained models on demand. Installation is lightweight by default (lite mode), with optional dependencies installed on first use. Configuration is managed through a JSON file that lets you customize paths, environment variables, and module search directories.

Use it for:

• Build video processing pipelines that apply classifiers or object detectors to frame sequences without writing custom inference code.
• Manage and download pre-trained ML models from ETA's registry and run inference on images or video frames.
• Extract and manipulate image embeddings, metadata, and annotations for downstream ML tasks.
• Automate computer vision workflows via the CLI for batch processing, model evaluation, or remote storage integration.
• Extend ETA with custom modules, pipelines, or models by registering them in the config and module search paths.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

ETA is an extensible computer vision and machine learning analytics toolkit that provides core utilities for working with images, videos, embeddings, and ML inference pipelines, along with a CLI for building and running analytics workflows.

Yes. ETA is actively maintained (release 4 days old), has low install friction, and is permissively licensed. It fills a genuine need for a modular, extensible analytics framework rather than a monolithic library. The 20 runtime dependencies are standard (numpy, opencv, scikit-image, Pillow) and well-maintained. No known security vulnerabilities. Best suited for teams building custom computer vision or ML analytics workflows; less ideal if you only need a single, narrow task (e.g., image resizing).

Install

voxel51-eta on PyPI

pip install voxel51-eta

uv add voxel51-eta

poetry add voxel51-eta

Installing voxel51-eta

Before you install

Low friction installation via pip with a pure-Python wheel. Active maintenance with a release 4 days old. The package pulls in 20 runtime dependencies including numpy, opencv-python-headless, and scikit-image; optional features like ffmpeg and TensorFlow are prompted on demand rather than bundled.

License in practice

Apache License (permissive) means you can use, modify, and distribute ETA freely in commercial and private projects, provided you include a copy of the license and note any changes you make.

Quickstart

pip install voxel51-eta

import eta.core

# Create a copy of config-example.json as eta/config.json to enable advanced features
# Then use eta CLI or import eta.core submodules for image/video utilities

Requires Python 3.10 or later. Advanced features (pipelines, model management) require a properly configured eta/config.json file. Optional dependencies like ffmpeg or TensorFlow must be installed separately if needed.

Verify before relying

• Whether the 20 runtime dependencies are all required for a minimal installation or if some are optional/conditional.
• Specific performance characteristics or scalability limits for large video or image datasets.
• Compatibility details between ETA and different TensorFlow versions (1.X vs 2.X) in practice.
• Whether the CLI is fully functional immediately after pip install or requires additional setup beyond config.json.

Package facts

Evidence: voxel51_eta-0.17.0-py3-none-any.whl

Tags

More Artificial Intelligence packages