This is a collection of Python utility packages for use with the IQR Vision system. Although much of the code is specific to IQR Lab's setup, the code is generally useful for similar Realsense camera + edge node systems (see more details here).
-
Install GNU Parallel
gnu-parallelon your computer and all the edge nodes:sudo apt-get install parallel
-
Create a
.envin your root directory following.env.example. -
Ensure that you can
sshinto each specificied host in theHOSTSvariable with the following command format:ssh [hostname]
SSH Config Details
Host iqr-vision-1 HostName iqr-vision-1.local User lab IdentityFile ~/.ssh/iqr-vision-1 -
Clone this repository and run:
pip install -e . -
Each host should have
librealsense2properly installed and have a camera connected. You can test this with the following command:iqr-vision testor if you prefer to use
runpypython -m iqr_vision.cli testExpected output for each host:
Connected devices: 1) [USB] Intel RealSense D455 s/n [...], update serial number: [...], firmware version: [...]
This module is specifically used for rgb color related functions. See the relevant function documentation for more details.
You must first install the iqr-multivideo executable to the edge nodes. Please do this through the cli:
iqr-vision install multivideoNote: Ensure that you call multivideo.close() after multivideo.start(). The start function is running an executable under the hood, which will terminate eventually on its own after the required duration, but subsequent runs will fail as the camera stream is already occupied. This is avoided by the pre_stop: bool = True parameter on the multivideo.start()
Example usage:
from pathlib import Path
from iqr_vision import multivideo
def record_timestamps():
import time
timestamps: list[str] = []
before_loop = time.time()
while time.time() - before_loop < 10: # Record for 10 seconds
nano = time.time_ns()
timestamps.append(
f"{nano // 1_000_000_000}.{str(nano % 1_000_000_000).zfill(9)}"
)
time.sleep(0.05) # Record timestamps for every 50ms
return timestamps
pids = multivideo.start(duration=60, bag_path="~/test_dir/test.bag")
timestamps = record_timestamps()
multivideo.stop(pids)
multivideo.transfer(local_bag_dir="bags/", remote_bag_path="~/test_dir/test.bag")
multivideo.docker_filter_rosbag(
local_bag_dir="bags/",
)
for bag_path in Path("bags/filtered_bags").glob("*.bag"):
multivideo.save_rosbag_images(
local_bag_path=bag_path,
destination_dir=f"images/{bag_path.stem}/",
timestamps=timestamps,
)This is a relatively straightforward Python package, proceed with standard practices. Document functions with Google format Python docstrings and format code with black.