Welcome to the AIND Video Encoding Benchmarks project documentation!

Aind.Behavior.VideoEncodingBenchmarks

A repository with code for benchmarking online video acquisition/encoding pipelines

Getting started

The easiest way to get started is to clone this repository and run the deploy.cmd script. This script will install the necessary dependencies. If you use any python-dependent scripts (e.g. generate valid schema json files or use the launcher script), make sure to run them from an activated environment (./.venv directory in the root of the repo).

To run the benchmark workflow you will need a valid set of schemas. Examples of how to generate these can be found in the ./examples/examples.py file. For the most part, you will only need to generate the rig schema, as the others can be automatically generated using the Launcher.cmd script. In summary:

  1. Generate the rig and task logic schemas by adapting the script in ./examples/examples.py. You should make a copy of the example and modify it outside the repository directory, otherwise, any untracked changes / new files in the local repository will flag it as “dirty” and the benchmarks will not be allowed to run.

  2. Once the schema is defined in the .py file, run the script to generate a valid .json file that will be the input to the benchmark workflow.

  3. Copy and Paste the generated rig schema (i.e. the .json file) to the target config folder (by default \\allen\aind\scratch\AindBehavior.db\AindVideoEncodingBenchmarks\Rig\<COMPUTERNAME>\<RIG_FILE>.json) and the task logic schema to (...)\AindVideoEncodingBenchmarks\TaskLogic\<TASK_LOGIC_FILE>.json

  4. Run the ./scripts/Launcher.cmd, from the root of the repository, and follow the prompt.

  5. Once Bonsai is running, double-click the UserInterface operator to open the GUI

  6. Click Start (Stop) to start (stop) the benchmark workflow.

Alternatively, you can run the main.bonsai script directly using the following command from the root of the repository:

"./bonsai/bonsai.exe" "./src/main.bonsai" -p TaskLogicPath="<PATH_TO_TASK_LOGIC.json>" -p RigPath="<PATH_TO_RIG.json>" -p SessionPath="<PATH_TO_SESSION.json>"

where <PATH_TO_TASK_LOGIC.json>, <PATH_TO_RIG.json>, and <PATH_TO_SESSION.json> are the paths to the task logic, rig, and session schemas, respectively.

General instructions

This repository follows the project structure laid out in the Aind.Behavior.Services repository.

Deployment

Deployment instructions can be found here.

Specifically, Ffmpeg and CUDA support must be installed on the computer.

Prerequisites

Pre-requisites for running the project can be found here.

Regenerating schemas

Instructions for regenerating schemas can be found here.

Contents:

Indices and tables