基于Raspberry Pi 3与OpenCV的网络摄像头流人脸识别问题咨询

Hey there! Let's dig into why your Raspberry Pi 3 camera stream is dying after a few minutes. I’ve dealt with similar flaky stream issues before, so here are targeted fixes to get that stable, long-running stream matching the test you’re aiming for:

FIFOs are finicky if read/write speeds don’t sync up, which can cause crashes after a few minutes:

• When creating your FIFO, use mkfifo -m 666 fifo264 to ensure proper read/write permissions for all processes.
• In your raspivid command, add -b 1500000 to lock in a consistent bitrate, and -n to disable the preview window (saves precious CPU/RAM). For example:

  raspivid -o fifo264 -t 0 -n -b 1500000 -fps 30
  

• In your Python script, avoid blocking indefinitely on FIFO reads. Add a timeout or check for data availability before reading to prevent the script from hanging when the buffer underflows/overflows.

The Pi 3’s CPU and RAM can get overwhelmed over time, leading to stream crashes:

• Use htop to spot background processes eating resources—close unused apps like VNC, Bluetooth, or web servers if you don’t need them.
• Check for overheating with vcgencmd measure_temp. If temps climb above 80°C, add a heatsink or improve ventilation (throttling kills stream stability fast).
• Disable swap if you’re not using it (swap can cause lag and crashes for real-time tasks) with sudo dphys-swapfile swapoff.

Dropped packets or weak WiFi signal are a common culprit for interrupted streams:

• Switch to a 5GHz WiFi network if your router supports it (less interference than 2.4GHz).
• Disable WiFi power management to stop the adapter from dozing off. Run sudo iwconfig wlan0 power off and add this line to /etc/rc.local (before exit 0) to make it permanent.
• Move the Pi closer to your router, or use a USB WiFi adapter with a better antenna if signal strength is low.

Your script might be accumulating memory or hitting unhandled errors over time:

• Use try/finally blocks to ensure file handles (like the FIFO) are closed properly, even if an error occurs. For example:

  fifo = open('fifo264', 'rb')
  try:
      # Your stream processing loop here
  finally:
      fifo.close()
  

• If you’re using OpenCV or frame processing, release each frame after use with frame.release() to avoid memory leaks.
• Add logging to your script (write errors or status updates to a file) so you can see exactly what’s failing when the stream crashes.

Rule out issues with your current setup by running a simplified stream first:

• On the Pi: raspivid -o - -t 0 -n -b 1500000 | nc <your-receiver-ip> <port>
• On the receiving device: nc -l <port> | mplayer -fps 30 -cache 1024 -
  If this runs stable for hours, the problem is isolated to your FIFO/Python script setup—we can tweak that further once you confirm.

If you’re still hitting issues, share a snippet of your Python script and the exact raspivid command you’re using, and we can dive deeper into the code!

内容的提问来源于stack exchange，提问作者user2355051