v2.1.1

VisionServer v2.1.1

This is the last version of the library to depend on WPILib 2023. The next release will be updated for WPILib 2024. This version of the code (and any updates will reside on the vs2-wpilib2023 branch of the repo for future reference.

What's Changed

• fix doxygen build action by @S1ink in #6
• patch apt-get by @S1ink in #8
• patch apt-get by @S1ink in #9
• patch apt-get (from main) by @S1ink in #10
• merge apriltag changes by @S1ink in #11
• Update to WPILib 2023 by @github-actions in #7
• fix ntables, update neon by @S1ink in #12
• Merge new deps by @github-actions in #13
• split up aprilpose routine for easier extendability by @S1ink in #14
• More calibration features by @S1ink in #15
• changes!? by @S1ink in #16

Full Changelog: v2.1.0...v2.1.1

v2.1.0

VisionServer v2.1.0

• Include OpenCV Aurco module for AprilTag support --> field pose estimator base pipeline
• Update Actions --> docs generation and update system
• New init() and close() pipeline methods
• Rework FPS capping (now "per-pipeline", adjustable over NT)
• More NEON optimized functions
• Remove unnecessary clutter

Full Changelog: v2.0.0...v2.1.0

MRI Vision

The robot code used during MRI 2022 still uses the VS1 interface, so the source for the entire program still exists on this repo.
Not much has changed since v1.1.0-Final other than the addition of a targeting overlay for the hub pipeline. A prebuilt binary is linked below for direct deployment.

VisionServer 2.0 Initial

VisionServer 2.0

• Converted to library format (.a and .so)
• Moved legacy code to separate directory
• New ::vs2 namespace
• Fully static API
• Multithreaded pipeline instancing
• Unlimited* pipelines and output streams
• Tensorflow Lite and EdgeTPU support (Axon)
• New and improved pipeline and target base classes

What's Changed

• prepare for multi-instance processing by @S1ink in #2
• Merge for v2.0.0 by @S1ink in #3
• merge new filestructure by @S1ink in #4

Full Changelog: v.0.3.0...v2.0.0

Final release of VS1

This release marks the end of VisionServer v1.x.x, and is basically pointless in practice. VS2.0 will succeed with support for multithreading processing instances and a much more robust internal architecture. This repo will also be revamped as to serve as more of library, so this is the last version that contains the entire project.
A build of the RR-season vision code can be found below. It has been updated to flush NT values every frame (this was a bug that existed in the previous release). This should be used when a more robust NT connection or compatibility with the old system is needed.

v1.0.0-beta

Just some cleaning up, better documentation, and small fixes.
Check out this guide for setting up the project.

Full Changelog: v1.0.0-alpha...v1.0.0-beta

v1.0.0-alpha

Official V1.0!

- Still titled alpha because stability has not been tested extensively

Feature Update V3

Full Changelog: v.0.2.0...v.0.3.0

Features:

• Creates a variable number of cameras based on the config file (/boot/frc.json)
• Camera settings (exposure, brightness, whitebalance) are automatically added to networktables
• 'VisionServer' framework for managing cameras and pipelines during runtime
• Networktable settings for switching cameras and pipelines (up to 3 currently supported)
• Option to show thresholding to help in debugging pipelines
• Extendable 'PipelineBase' class for creating new pipelines
• Set compression (although this cannot be changed during runtime)
• Basic threading support for VisionServer (only 1 pipeline can currently be used)

Demo:

See the README

more features

v.0.2.0

some reorganization and support for switching up to 3 pipelines durin…

Initial Release

v.0.1.0

SO MANY IMPROVEMENTS