Depth Logging
Logging detections with depth is performed in mostly the same way as regular visual detection logging, except the source of the video feed is different.
Logging visual detections with depth is done by performing the following steps:
- Start whichever ROS node used to create point clouds in the ros::PointCloud2 format. Ensure the point clouds are being published to the topic
/stereo/points2 - Start YOLO with
roslaunch darknet_ros darknet_ros.launch - Start the image builder with
rosrun image_draw image_draw_node - Start the logger with
rosrun detection_logger visual_logger
The results of the detections will still be saved in a folder titled with the current year in the working directory.
A copy of the detected image will be saved in a folder called "images" in the subdirectory of your working directory, under the current date. Details of the images and detections will be saved in the same subdirectory as "images", but in a folder titled "text" and file titled "output.txt".
"output.txt" is a file containing a comma-separated value list of each detection. The entries are structured in the following way:
detection class, confidence, timestamp, x-coordinate of bounding box upper-left corner, y-coordinate of bounding box upper-left corner, bounding box width, bounding box height, distance to detected object, file name of the image containing the bounding box.