Welcome to the Open Media Processing Framework (OpenMPF) Evaluation Framework Project!
OpenMPF provides a platform to perform content detection and extraction on bulk multimedia, enabling users to analyze, search, and share information through the extraction of objects, keywords, thumbnails, and other contextual data.
OpenMPF enables users to build configurable media processing pipelines, enabling the rapid development and deployment of analytic algorithms and large-scale media processing applications.
Simplify large-scale media processing and enable the extraction of meaningful content
Apply cutting-edge algorithms such as face detection and object classification
Integrate into your existing environment or use OpenMPF as a standalone application
This repository contains code for the OpenMPF evaluation framework and related files. The framework is used for generating metrics such as speed, accuracy, precision, recall, etc., of OpenMPF component algorithms, depending on the algorithm type (e.g. face detection, text detection, speech detection, object detection and classification, etc.).
- Parent OpenMPF Project
- OpenMPF Core
- Components
- Component APIs:
- OpenMPF Build Tools
- OpenMPF Docker
- OpenMPF Web Site Source
- OpenMPF Evaluation ( You are here )
The evaluation framework uses the CLI Runner to execute jobs using pre-built OpenMPF Docker images, such as those on Docker Hub.
To run the evaluation framework run:
python3 evaluation_framework.py <docker-image-name-with-registry-and-tag> <path-to-media-file>
For example, to run OCV face detection on the sample image file, run:
python3 evaluation_framework.py openmpf_ocv_face_detection:latest /home/mpf/openmpf-projects/openmpf-evaluation/data/meds-af-S419-01_40deg.jpg
For more information about OpenMPF, including documentation, guides, and other material, visit our website.
For a latest snapshot of what tasks are being worked on, what's available to pick up, and where the project stands as a whole, check out our workboard.