ichiro-its / ninshiki Goto Github PK

Change this package name into ninshiki_tensorflow as it only implements TensorFlow for detection.

Use a ROS 2 subscription to subscribe on image or compressed image topic from the Shisen project, and then show the received image on windows gui using OpenCV's imshow() as in the Shisen's viewer (but in Python).

update detector.py and viewer.py according to shisen

Publish the detection result on <node_name>/detections topic. The detection result probably will require a new Msg to store the message (it's better to put the new Msg on separate package).

Use libraries/frameworks like keras, TensorFlow, etc. to detect object from the received image using pretrained model. The detected object nor the image do not have to be shown on a new window, outputing the detected object result on the terminal is enough.

Add a GitHub workflows to build and test this project using ROS 2 CI.