V4L2-compliant webcams that support a relatively recent Linux kernel version should be fine. When you say “the Jetson”, what I think you mean is will it work with application XYZ? That is dependent on the particular application if they support the color format that the camera outputs. Typically USB cameras that output RGB or YUV video are widely supported (I think C270 does).
There have been some instances where folks trying to get USB webcams working with machine learning applications, where those cameras only support compressed JPEG or H.264/H.265 streaming. In that case the application may need to support the decompression in it’s pipeline, that’s why the raw RGB/YUV cameras typically work easier across a broad selection of sw.
The latency of USB cameras is often higher than that of directly connected CSI cameras, and supported throughput and/or resolution is often lower.
That being said, any USB camera that works in Linux should work with most applications on the Nano, assuming you can configure the appropriate camera device as input.
For the aforementioned Logitech C270 types of cameras how relative large is the latency. I am running the image detection on a partially dynamic environment (Environment movement is less] but the mount of the camera is moving. Is a CSI camera a better option ?
Yes, as I said, CSI cameras are generally a better option if you want low latency.
There are multiple reasons for this, including that “webcams” often run at low frame rates (like 30 fps) and often run at USB 2.0 speeds, which cause transmission of each frame to itself take a lot of time.
I have generally found less buffering and latency in CSI camera setups, although of course the specifics still matter.
Go for lower latency and higher frame rate to reduce latency, btw. Trying to do real-time analytics with a moving camera that captures 4k at 30 Hz is not helpful, both because of latency, and because of the excessive image skew you get from the rolling shutter during that time. Aim for 120 Hz if you can. Frame rate is more important than resolution, most of the time!
Hi
Please click the below link to know the camera options for Jetson Nano from e-con Systems.
e-con Systems, the NVIDIA Jetson Elite partner is a pioneer in delivering high-quality cameras for the NVIDIA® Jetson™ platform. Backed by the world’s leading embedded AI computing platform, we regularly launch many MIPI CSI-2 Jetson cameras that guarantee high performance while ensuring low-power consumption.