Common Object Detection Problems
Understanding common failure modes helps you debug and improve your detection systems.
6 min•By Priygop Team•Updated 2026
Troubleshooting Detection Issues
- Too many false positives: lower confidence threshold is too permissive — raise it (try 0.6-0.7)
- Missing objects: confidence threshold too high — lower it, or check if the object class is in the model's vocabulary
- Duplicate boxes: NMS threshold too high — lower the IoU threshold for NMS (try 0.3-0.4)
- Poor accuracy in low light: pre-process image — increase brightness, apply CLAHE
- Correct class, wrong location: bounding box coordinates are off — check if format is (x,y,w,h) vs (x1,y1,x2,y2)
- Very small objects missed: use a model with higher input resolution, or crop and detect sub-regions
- Objects outside training classes: pretrained models only detect classes they were trained on — you need custom training for other classes
Diagram
Loading diagram…
Machine Learning follows a structured pipeline from data to deployment
Key Takeaways
- Understanding common failure modes helps you debug and improve your detection systems.
- Too many false positives: lower confidence threshold is too permissive — raise it (try 0.6-0.7)
- Missing objects: confidence threshold too high — lower it, or check if the object class is in the model's vocabulary
- Duplicate boxes: NMS threshold too high — lower the IoU threshold for NMS (try 0.3-0.4)