System design
System design guides
Putting a station together: PCs, triggers, cables, mounting and the parts list for a working system.
System design · Troubleshooting
Anomaly detection early stopping with high training loss
Diagnose anomaly detection early stopping with high loss, choose the right validation metric, check data leakage, and verify defects on held-out images.
System design · Explainer
Visual servoing camera integration: adapting image input
Validate visual servoing camera integration with checks for authorized acquisition, pixel format, calibration, depth, frame freshness and latency.
System design · Troubleshooting
Vision picking project changeover: diagnose ownership faults
Diagnose vision picking project changeover faults with stop acknowledgments, ownership checks and safe, documented recovery procedures.
System design · Troubleshooting
Unknown pixel format: diagnose missing depth streams
Diagnose unknown pixel format errors when color works but depth is missing, with stream checks, driver evidence and supported software validation.
System design · How-to
Reset camera settings programmatically before capture
Reset camera settings programmatically before capture using a default user set, an explicit load command, and verified exposure readback.
System design · Troubleshooting
Multi-camera acquisition freezes: callbacks and buffer queues
Diagnose multi-camera acquisition freezes by tracing trigger edges, callbacks, buffer returns and queue drops, then build matched calibration sets.
System design · Troubleshooting
Machine vision network congestion: diagnose transfer faults
Diagnose machine vision network congestion with bandwidth estimates, switch counters and timed tests that separate link faults from endpoint failures.
System design · Troubleshooting
Machine vision callback disconnect after idle periods
Diagnose a machine vision callback disconnect after idle with timestamped logs, controlled tests, and supported lifetime or reconnection fixes.
System design · Explainer
Local vision recipe switching from operator buttons
Design local vision recipe switching with documented interfaces, controlled stop/load sequencing, readiness checks, timeouts and recovery.
System design · Explainer
Live feed circle detection with a circle Hough transform
Live feed circle detection using a circle Hough transform: set radius bounds, control edge contrast, reject false circles and measure latency.
System design · Troubleshooting
GenTL camera open failure after successful discovery
Diagnose GenTL camera open failure by refreshing discovery lists, tracking handle lifetimes, and separating open errors from cleanup faults.
System design · Troubleshooting
Conveyor vision startup synchronization troubleshooting
Diagnose conveyor vision startup synchronization warnings with paired startup tests, encoder checks, trigger timing, filters, and readiness checks.
System design · Troubleshooting
Conveyor vision scene timing: troubleshoot stale pick data
Trace conveyor vision scene timing through triggers, clocks, encoder counts and controller queues to diagnose stale pick data and verify recovery.
System design · Troubleshooting
Contour position jitter under automatic exposure
Diagnose contour position jitter by separating exposure changes, light flicker, noise and pixel quantization, then verify stable measurements.
System design · Explainer
Camera calibration target flatness: when paper is enough
Assess camera calibration target flatness, printing error and corner detection, then choose mounted paper or a rigid target with independent checks.
System design · Explainer
Barcode verification before picking with multiple robots
Design barcode verification before picking with item IDs, encoder tracking, scan deadlines and robot handshakes that reject stale or ambiguous results.
System design · Explainer
Automatic iris, exposure and gain control in machine vision
Understand automatic iris, exposure and gain control, check camera and lens support, and set limits for blur, depth of field and noise.
System design · Explainer
Visual servoing loop design for eye-in-hand picking
Design a visual servoing loop for eye-in-hand picking with detection, tracking, calibration, timing checks, bounded motion and target-loss handling.
System design · Guide
Sorting conveyor camera selection: pixels, speed and blur
Sorting conveyor camera selection with worked calculations for pixels, belt-speed exposure limits, capture rate, lighting and prototype trade-offs.
System design · Checklist
Machine vision software upgrade validation checklist
Plan a machine vision software upgrade with backups, compatibility checks, encoder records and measurable conveyor-tracking acceptance tests.
System design · Troubleshooting
Linux GigE Vision camera discovery troubleshooting
Diagnose Linux GigE Vision camera discovery when ping works, separating network paths, GenTL loading, transport services and control ownership.
System design · Explainer
Line scan image overlap across frame boundaries
Build line scan image overlap in application memory: size retained rows, separate output height from stride, and prevent missed or duplicate detections.
System design · Troubleshooting
Intermittent camera acquisition failure troubleshooting
Diagnose camera acquisition failure by isolating triggers, power, cabling, interference and host faults, then verify calibration and recovery.
System design · Troubleshooting
Image sensor not detected: check camera module connections
Diagnose an image sensor not detected error with safe connector checks, power and interface tests, and a minimal acquisition recovery check.
System design · Explainer
Visual servoing camera integration: adapting image input
Validate visual servoing camera integration with checks for authorized acquisition, pixel format, calibration, depth, frame freshness and latency.
System design · Troubleshooting
Vision picking project changeover: diagnose ownership faults
Diagnose vision picking project changeover faults with stop acknowledgments, ownership checks and safe, documented recovery procedures.
System design · Troubleshooting
Unknown pixel format: diagnose missing depth streams
Diagnose unknown pixel format errors when color works but depth is missing, with stream checks, driver evidence and supported software validation.
System design · How-to
Reset camera settings programmatically before capture
Reset camera settings programmatically before capture using a default user set, an explicit load command, and verified exposure readback.
System design · Troubleshooting
Multi-camera acquisition freezes: callbacks and buffer queues
Diagnose multi-camera acquisition freezes by tracing trigger edges, callbacks, buffer returns and queue drops, then build matched calibration sets.
System design · Troubleshooting
Machine vision network congestion: diagnose transfer faults
Diagnose machine vision network congestion with bandwidth estimates, switch counters and timed tests that separate link faults from endpoint failures.
System design · Troubleshooting
Machine vision callback disconnect after idle periods
Diagnose a machine vision callback disconnect after idle with timestamped logs, controlled tests, and supported lifetime or reconnection fixes.
System design · Explainer
Local vision recipe switching from operator buttons
Design local vision recipe switching with documented interfaces, controlled stop/load sequencing, readiness checks, timeouts and recovery.
System design · Explainer
Live feed circle detection with a circle Hough transform
Live feed circle detection using a circle Hough transform: set radius bounds, control edge contrast, reject false circles and measure latency.
System design · Troubleshooting
GenTL camera open failure after successful discovery
Diagnose GenTL camera open failure by refreshing discovery lists, tracking handle lifetimes, and separating open errors from cleanup faults.
System design · Troubleshooting
Conveyor vision startup synchronization troubleshooting
Diagnose conveyor vision startup synchronization warnings with paired startup tests, encoder checks, trigger timing, filters, and readiness checks.
System design · Troubleshooting
Conveyor vision scene timing: troubleshoot stale pick data
Trace conveyor vision scene timing through triggers, clocks, encoder counts and controller queues to diagnose stale pick data and verify recovery.
System design · Troubleshooting
Contour position jitter under automatic exposure
Diagnose contour position jitter by separating exposure changes, light flicker, noise and pixel quantization, then verify stable measurements.
System design · Explainer
Camera calibration target flatness: when paper is enough
Assess camera calibration target flatness, printing error and corner detection, then choose mounted paper or a rigid target with independent checks.
System design · Explainer
Barcode verification before picking with multiple robots
Design barcode verification before picking with item IDs, encoder tracking, scan deadlines and robot handshakes that reject stale or ambiguous results.
System design · Explainer
Automatic iris, exposure and gain control in machine vision
Understand automatic iris, exposure and gain control, check camera and lens support, and set limits for blur, depth of field and noise.
System design · Explainer
Visual servoing loop design for eye-in-hand picking
Design a visual servoing loop for eye-in-hand picking with detection, tracking, calibration, timing checks, bounded motion and target-loss handling.
System design · Guide
Sorting conveyor camera selection: pixels, speed and blur
Sorting conveyor camera selection with worked calculations for pixels, belt-speed exposure limits, capture rate, lighting and prototype trade-offs.
System design · Checklist
Machine vision software upgrade validation checklist
Plan a machine vision software upgrade with backups, compatibility checks, encoder records and measurable conveyor-tracking acceptance tests.
System design · Troubleshooting
Linux GigE Vision camera discovery troubleshooting
Diagnose Linux GigE Vision camera discovery when ping works, separating network paths, GenTL loading, transport services and control ownership.
System design · Explainer
Line scan image overlap across frame boundaries
Build line scan image overlap in application memory: size retained rows, separate output height from stride, and prevent missed or duplicate detections.
System design · Troubleshooting
Intermittent camera acquisition failure troubleshooting
Diagnose camera acquisition failure by isolating triggers, power, cabling, interference and host faults, then verify calibration and recovery.
System design · Troubleshooting
Image sensor not detected: check camera module connections
Diagnose an image sensor not detected error with safe connector checks, power and interface tests, and a minimal acquisition recovery check.
Shop: Configure a system · All guides