Circum is a set of tools for detecting and tracking moving objectsvia a variety of distributed sensors.
Circum is a distributed, multi sensor fusion system for detecting and tracking people. It applies techniques similar to systems developed for autonomous vehicles to detect and track moving objects (DATMO). Circum uses late fusion, meaning that detections are classified per sensor and then fused (associated and deduplicated) and tracked after. Because different sensors provide different capabilities (e.g. point vs volume detection), these properties will be combined in the final tracking output.
Circum is intended for art installations wanting to use human presence as an input into an interactive installation.
pip3 install circum
Usage: circum [OPTIONS] Options: -n, --name TEXT The service name [required] -i, --interface TEXT The interface to bind to. -p, --port INTEGER The port to bind to. -e, --endpoint TEXT Names of endpoints to connect to. Can be specified multiple times. If no endpoints are specified, all available endpoints will be used. --help Show this message and exit.
Usage: circum-endpoint [OPTIONS] COMMAND [ARGS]... Options: --name TEXT The service name --interface TEXT The interface to bind to. --port INTEGER The port to bind to. --pose FLOAT... The pose of the sensor. Expressed in x y z yaw(Rx) pitch(Ry) roll(Rz) order. Units are meters and degrees. +Z is the direction of sensor view. X & Y follow the right hand rule. If a pose provider is installed, this will override it. --pose-provider [<available providers installed from plugins will be listed here>] The pose provider to use for automatically determining the sensor pose. NOTE: this is currently unsupported --help Show this message and exit. Commands: simulator <additional sensors installed from plugins will be listed here>
To get the help for a particular sensor add --help after the appropriate command:
?> circum-endpiont simulator --help Usage: circum-endpoint simulator [OPTIONS] Options: --update_interval FLOAT Rate to send updates. --num_objects INTEGER Number of objects to simulate --help Show this message and exit.
The circum service can be started with unique name and either a list of endpoints to connect to or it will find and connect to all circum endpoints on a network. Once connected to the endpoints, it will use pose and field of view information from each endpoint to combine tracking data into a single view which it will then transmit to clients whenever updated.
Circum, by convention, uses a right handed coordinate system with y as the vertical axis and +y zenith pointing. Circum, itself, doesn't enforce or care about this, but, if a given application is deviating from this convention, it is important to ensure that the endpoints are all using the same coordinate convention or the resulting tracks will be unusable.
The circum service will advertise itself via zeroconf service discovery. It will advertise under
Endpoints perform detection and classification and transmit information about the detected objects to the core service. At the very least, the endpoint must transmit a centroid of a detected person. The core service operates on this. Any additional information is added into the fused track for clients to consume. Each endpoint is configured with a unique name.
Circum supports automatically updating tracked objects based on the pose of the sensor. This assumes a right handed sensor coordinate system with +z extending out from the sensor, +y vertical, and +x horizontal when θx, θy, and θz are all 0. If the sensor has a different coordinate system, it mist be transformed to this system first.
Each endpoint is exposed as a discoverable zeroconf service. They are advertised under
Circum is distributed with a simulator endpoint. Additional endpoint sensor types are installed via plugins.
Sensors in Development:
The endpoints will advertise under
The type of endpoint will be noted in the service properties
After installing circum and downloading the git repo, run the following in separate terminals (you can substitute your own values for FOO, BAR, 8081, and 8082):
circum-endpoint --name FOO --port 8081 simulator circum-service --name BAR --port 8082 python3 .\examples\demo_client.py --service BAR
Circum would not have been possible without the following references:
- R. Omar Chavez-Garcia. Multiple Sensor Fusion for Detection, Classification and Tracking of MovingObjects in Driving Environments. Robotics [cs.RO]. Université de Grenoble, 2014. English. <tel-01082021>
- Sensor Fusion and Object Tracking using an Extended Kalman Filter Algorithm — Part 1 & Part 2
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size circum-0.0.6.tar.gz (16.2 kB)||File type Source||Python version None||Upload date||Hashes View|