Stuck #2
Description
.env file specifying environment variables for the YOLOv8 ROS2 package
WARNING: You must rebuild the yolov8 package after changing any of these variables so the .env file is placed in the package's share directory
Specifies the path to the root dir of your TensorRT install so FindTensorRT.cmake can find the libraries when building
TENSORRT_DIR=/usr/src/tensorrt/
TENSORRT_DIR=$ENV{HOME}/libs/TensorRT-8.6.1.6/
Nice level to run the inference node at. This is the priority of the process. The lower the number, the higher the priority (0-19) compared to other processes on the system.
NICE_LEVEL=7
The compute capability version (architecture) of your NVIDIA GPU. See Compute Capability" on https://developer.nvidia.com/cuda-gpusgpus for a list of compute capability versions for each GPU.
OpenCV will be built with this compute capability version as it builds against CUDA.
CUDA_ARCH_BIN=7.2
The OpenCV version to build
OPENCV_VERSION=4.5.4
Path to ONNX model for TensorRT engine to build from relative to the src/yolov8/models/ directory
ONNX_MODEL="/home/barelangfc1/bfc_ros2/src/yolov8/models/yolov8n-segbatch1.onnx"
Camera topics to subscribe and run inference on
CAMERA_TOPICS="/image_raw"
Hz the camera image buffer should be batched the model if the camera image buffer isn't full
(timeout if not all cameras have published an image)
CAMERA_BUFFER_HZ=30
Whether to visualize the masks on an image (as it may cost a small amount of performance)
VISUALIZE_MASKS=true
Whether or not to create a one channel segmentation mask from the original mask(s)
This is mainly for tasks like LiDAR projection where a one channel mask is needed
ENABLE_ONE_CHANNEL_MASK=true
Whether to visualize the one channel segmentation mask on an image (as it may cost a small amount of performance)
VISUALIZE_ONE_CHANNEL_MASK=true
Precision to be used for inference
PRECISION=FP16
Calibration data directory (must be specified when using INT8 precision)
CALIBRATION_DATA_DIRECTORY=""
Probability threshold used to filter detected objects
PROBABILITY_THRESHOLD=0.25
Non-maximum suppression threshold
NMS_THRESHOLD=0.65
Max number of detected objects to return
TOP_K=100
Segmentation config options (output dimensions of the ONNX model)
SEG_CHANNELS=16
SEG_H=640
SEG_W=640
SEGMENTATION_THRESHOLD=0.5
Class names
CLASS_NAMES=car
when i run this below commmand. I stuck on this below result. This above is my yolov8.env. What i need to do?
ros2 launch yolov8 yolov8.launch.py
[INFO] [launch]: All log files can be found below /home/barelangfc1/.ros/log/2024-12-16-12-37-18-140285-barelangfc1-desktop-47984
[INFO] [launch]: Default logging verbosity is set to INFO
['/image_raw']
YOLOv8 Parameters:
onnx_model_path: /home/barelangfc1/bfc_ros2/src/yolov8/models/yolov8n-segbatch1.onnx
model_dir: /home/barelangfc1/bfc_ros2/src/yolov8/models/yolov8n-segbatch1.onnx
camera_topics: ['/image_raw']
camera_buffer_hz: 30.0
visualize_masks: True
enable_one_channel_mask: True
visualize_one_channel_mask: True
nice_level: 7
precision: FP16
calibration_data_directory:
probability_threshold: 0.25
nms_threshold: 0.65
top_k: 100
seg_channels: 16
seg_h: 640
seg_w: 640
segmentation_threshold: 0.5
/home/barelangfc1/bfc_ros2/src/yolov8/models/yolov8n-segbatch1.onnx
[INFO] [ros_segmentation-1]: process started with pid [48080]
[ros_segmentation-1] [INFO] [1734327439.226465941] [rclcpp]: Creating YoloV8 engine --- Could take a while if Engine file is not already built and cached.
[ros_segmentation-1] [INFO] [1734327444.628841834] [rclcpp]: YoloV8 engine created and loaded into memory.
[ros_segmentation-1] [INFO] [1734327444.846227540] [yolo_v8]: Camera Topics:
[ros_segmentation-1] [INFO] [1734327444.846735288] [yolo_v8]: /image_raw