README.md

Gesture detection

Dataset

Checkout here

Dependencies:

• Tensorflow
• autoware_msgs

Running:

• Run YOLO2

$ roslaunch cv_tracker yolo2.launch

• Main code:

$ rosrun traffic_gesture_recognition ros_gesture_detection.py

• Filter:

$ rosrun traffic_gesture_recognition gesture_filter.py

• Show debug images:

$ rosrun image_view image_view image:=/police_gesture/image_overlay

• Toggle debug images:

$ rosservice call /set_debug "data: true"
$ rosservice call /set_debug "data: false"

• Toggle policeman detection

$ rosservice call /verify_policeman "data: true"
$ rosservice call /verify_policeman "data: false"

• Run from video_file

roslaunch traffic_gesture_recognition video_file.launch

• Set correct topic for launch (change videofile/image_raw to topic of your choice)

rosrun topic_tools relay /videofile/image_raw /image_raw