Forked from official YOLOv7 repo. See here for more information on detection variables.
- Install NVIDIA GPU drivers and CUDA Toolkit
- Install
requirements.txt
pip3 install -r requirements.txt
- Build package with
catkin_make
- Launch camera node, output
Image
message to/camera/image
topic. - Run
detector.py
rosrun yolov7_ros detector.py
- Stream should start and detections output to
/detections
topic asDetections
message type.
Header header
yolov7_ros/Detection[] dets
List of detections
string name
Name of object
int8 conf
Confidence of detection
int16[4] bbox
Position of bounding box top-left and bottom-right [x1,x2, y1,y2]
.