Best Practices#

This chapter summarizes proven recommendations for building robust, efficient, and maintainable pypylon applications in production environments.

Resource Management#

Always ensure that camera resources are acquired and released properly.

with pylon.InstantCamera(pylon.FirstFound) as camera:
    # use camera

Benefits:

• Automatic cleanup
• Exception safety
• Prevents resource leaks

Device Identification#

Never rely on device indices as they are not stable:

devices[0]

Use unique identifiers instead:

• Serial number
• User-defined name

See Discovering and Selecting a Camera for details.

パフォーマンスの最適化#

Use Appropriate Grab Strategy#

• LatestImageOnly → low latency
• OneByOne → complete processing

Reduce Data at Source#

• Specify a ROI.
• Reduce resolution.
• Lower frame rate if possible.

Avoid Unnecessary Copies#

• Copy only when required.
• Reuse buffers when possible.

Threading Design#

Separate Acquisition and Processing#

Camera → Queue → Processing

Benefits:

• Better CPU utilization
• Avoids blocking acquisition

Use Producer–Consumer Pattern#

• Acquisition thread pushes images.
• Processing threads consume them.

Use Thread-Safe Queues#

import queue
q = queue.Queue()

Multi-Stage Pipelines#

Structure complex applications as pipelines:

Grab → Preprocess → Analyze → Output

Guidelines:

• Each stage has one responsibility.
• Communicate via queues.
• Allow independent scaling per stage.

Error Handling#

Always Check Grab Success#

if grab_result.GrabSucceeded():
    process(grab_result.Array)
else:
    log_error(grab_result.ErrorDescription)

Catch and Log Exceptions#

try:
    # acquisition code
    pass
except Exception as e:
    log_error(e)

Implement Recovery Logic#

Error → StopGrabbing → Retry → Continue

Shutdown Strategy#

Implement clean shutdown mechanisms:

• Use threading.Event.
• Avoid blocking calls without timeout.
• Stop acquisition explicitly.

stop_event.set()
camera.StopGrabbing()

OpenCV Integration Rules#

• Run imshow() only in main thread.
• Avoid GUI in worker threads.

System Monitoring#

Track key metrics:

• Frame rate
• Processing latency
• Queue sizes
• CPU usage

Configuration Management#

• Store camera parameters externally.
• Avoid hardcoded values.
• Validate configuration at startup.

Recommendations for Testing#

• Test with real hardware.
• Simulate load conditions.
• Verify recovery behavior.

Common Pitfalls#

• Using image buffers after release
• Blocking queues indefinitely
• Ignoring synchronization requirements
• Mixing UI and worker threads

Mental Model#

Reliable system = correct architecture + error handling + performance tuning

Key Takeaways#

• Design for robustness from the start.
• Separate concerns (acquisition vs. processing).
• Validate every external interaction.
• Plan for failure and recovery.