dive

DIVE is a web interface for performing data management, video annotation, and running a portion of the algorithms stored within the VIAME repository. When compiled, docker instances for DIVE can be run either as local servers or online in web services. A sample instance of DIVE is running on a public server at viame.kitware.com.

Features

• video annotation
• still image (and image sequence) annotation
• deep integration with VIAME computer vision analysis tools
• single-frame boxes, polygons, and lines
• multi-frame bounding box tracks with interpolation
• Automatic transcoding to support most video formats
• Customizable labeling with text, numeric, multiple-choice attributes

Documentation

• Client User Guide
• Client Development Docs
• Server Development Docs
• Deployment Overview
• Running with Docker Compose
• Upgrading to Girder 5

Technologies Used

DIVE uses Girder 5 for data management and has a typical Girder + Girder Worker + Docker architecture (MongoDB, RabbitMQ, and Redis). See Running with Docker Compose and Upgrading to Girder 5 for deployment details.

• The client application is a standard @vue/cli application.
• The job runner is built on celery and Girder Worker. Command-line executables for VIAME and FFmpeg are built inside the worker docker image.

Docker Compose profiles

DIVE supports two Docker Compose modes:

• Default mode (GPU-enabled): starts the web stack, the standard worker, and GPU workers (girder_worker_pipelines, girder_worker_training) for VIAME pipelines and training.
• CPU-only mode (--profile cpu): starts only the standard worker (girder_worker_default) and omits GPU workers.

Commands

# Default mode (GPU-enabled workers)
docker compose up -d

# CPU-only mode
docker compose --profile cpu up -d

When GPU workers are not running (for example, in CPU-only mode), the web UI and API automatically disable pipeline and training actions.

Example Data

Input

DIVE takes two different kinds of input data, either a video file (e.g. .mpg) or an image sequence. Both types can be optionally accompanied with a CSV file containing video annotations. Example input sequences are available at https://viame.kitware.com/girder#collections.

Output

When running an algorithmic pipelines or performing manual video annotation (and saving the annotations with the save button) output CSV files are produced containing output detections. Simultaneously a detection plot of results is shown underneath each video sequence.