Welcome to VIAME

In cooperation with the National Oceanic and Atmospheric Administration’s (NOAA) Automated Image Analysis Strategic Initiative (AIASI), Kitware and its partners have developed Video and Image Analytics for Marine Environments (VIAME). VIAME is an open source computer vision software platform designed for do-it-yourself artificial intelligence (AI). It is an evolving toolkit that contains many workflows used to generate different object detectors, full-frame classifiers, image mosaics, rapid model generation, image and video search, and methods for stereo measurement.

Originally targeting marine species analytics, it now contains many common algorithms and libraries, and is also useful as a generic computer vision library outside of underwater image and video. VIAME is available as a desktop or a web application.

Install VIAME
Support & Services
Documentation
Source Code

VIAME in Action

VIAME offers three (3) core detection training workflows dependent on user needs and available training data.

Use Kitware’s Interactive Query Refinement (IQR) to rapidly generate models for new object classes with very little user effort.

Feel the freedom to use VIAME web interface for basic object detection and annotation or download the desktop version for higher level analytics.

Access VIAME’s specialized object detectors and functionalities specific to maritime and other domains of interest, including multiple general purpose detectors for wide applicability.

VIAME users have access to multiple trackers that use different techniques, such as deep learning addressing multiple domains.

Poor video quality or collection environments can be improved using VIAME’s image enhancement features.

VIAME’s image registration and mosaicing can be used for multi-camera and multi-modality experiments in multiple domains.

VIAME offers multiple types of annotation features down to the pixel-level to support object detection training workflows.

We continue to conduct research and development using scene semantics and 3D reconstruction to improve object classification.

VIAME uses knowledge-driven scene understanding using deep-learning techniques to segment a scene into object types with high accuracy.