Camera recommendations for computer vision projects

Projects & Demos
1

Hi everyone,

I’m working on computer vision projects using Python and Degirum, mainly on Raspberry Pi 5 with accelerators like Hailo-8L. I’m currently evaluating different cameras, from maker-level to industrial-grade.

I’d really appreciate recommendations considering:

  • Linux / Python / OpenCV / Degirum compatibility

  • Driver and SDK stability

  • Image quality (global shutter, FPS, latency)

  • Reliability for continuous operation

  • Price vs performance

  • Real production experience

Typical use cases:

  • Object detection

  • Tracking

  • Basic visual inspection

  • Edge deployment

Thanks in advance!

2

Well, it depends on your application really.
Answer for yourself the following questions:

  1. What WxH resolution do you need to fulfill your task? 720P? FullHD?
  2. In which environment will you use your camera: indoor with controlled temperature and humidity, or outdoor in mild climate, or in severe climate conditions?
  3. What connection interface do you see suitable for deployment: USB, Ethernet, WiFi, CSI?
  4. What FPS do you need for your application? Is it hard real time reaction which requires 30+ FPS (robotics) or is it used occasionally and do not require high FPS?
  5. What should be the field of view (FOV)? Wide angle or telephoto?
  6. Do you need automatic focus or you need long focus range with fixed focus?
  7. Do you envision to work in varying lighting conditions (i.e. do you need aperture controls)?
  8. Do you need to point camera into different directions (have PTZ controls)?
  9. Do you need optical zoom?
3

I’m curious on this because I have used Raspberry Pi 5 with AI Hat 26TOPs https://plate.afrifanom.com

This was able to support 4 cameras streaming at 4M mega pixels per second. For a high grade, I had suggest changing the Raspberry Pi in itself since it’s GPU is virtually non existent.