0.5.0

Version 0.5.0

This major release expands the camera-traps application with a new functional mode, referred to as demo mode. Unlike the original simulation mode, when operated in demo mode, the software relies on a real camera mounted and accessible as a Linux device. It uses motion detection to drive the camera and generate new images.

Additionally, when running in demo mode, the software has support for detecting a specific object and notifying a downstream process via an MQTT message. In particular, this approach supports integrations with drones which can fly to a particular camera that has spotted an object of interest.

The major new components for this release include:

1. A new Image Detecting plugin, with Linux Motion software
   for detecting motion and generating new images from a mounted camera.
2. A new Detection Reporter plugin, which determines if the configured object of interest has been detected (up to the configure confidence threshold). Detection Reporter maintains a CSV file of all detections.
3. A new CKN-MQTT Camera Traps container, which reads the
   output of the Detection Reporter and sends an MQTT message to the CKN broker for each detection.

There were a number of additional major changes, including:

1. The installer has been substantially expanded to provide support for automatically installing all of the new components mentioned above. Additionally, the installer now recognizes a new configuration, mode: demo, to specify running in demo mode versus simulation mode.
2. A number of improvements have been mode to the image scoring plugin and associated machine learning techniques. Support for a number of new models has been added, including support for ultralytics-based models. There is also support for running an ICICLE Model Commons "inference server" as a separate HTTP server container, aiding with plug-and-play.
3. A number of quality of life improvements have been made, including reducing build times with better ways of updating file permissions in the container images.