IQR Vision Utilities

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).

Installation

Install GNU Parallel gnu-parallel on your computer and all the edge nodes:
```
sudo apt-get install parallel
```
Create a .env in your root directory following .env.example.

Ensure that you can ssh into each specificied host in the HOSTS variable 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 librealsense2 properly installed and have a camera connected. You can test this with the following command:
```
iqr-vision test
```
or if you prefer to use runpy
```
python -m iqr_vision.cli test
```
Expected output for each host:
```
Connected devices:
1) [USB] Intel RealSense D455 s/n [...], update serial number: [...], firmware version: [...]
```

Usage

`multivideo`

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 multivideo

Note: 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,
    )

Contributing

This is a relatively straightforward Python package, proceed with standard practices. Document functions with Google format Python docstrings and format code with black.