If your task displays no AI detection frames, use the following steps to troubleshoot and resolve the issue.
If your task stream shows images but no AI detection frames, check the current detection settings of your task:
In "System Settings" of the task, confirm the correct detection event is chosen and that it's of the object detection type. Some specialized detections, like bloackage detection or screen availability, don’t show detection frames as they are not object-based.
Check the "AI Confidence Threshold" in "System Settings" of the task.
Threshold -> 0%：If set to 0%, you'll see all detected objects.
The default system setting is 30%. Increasing the threshold to 50-70% can enhance precision but may yield fewer frames. Settings above 80% might result in hardly any frames.
Adjust the AI Confidence Threshold according to your preferred detection approach.
For more detailed instructions, refer to our documentation:
當遇到沒有物件偵測方框，也可能是AI 模型的問題，您可以將透過以下方法檢查。
If the AI model isn’t activated, your task will show an alert.
Check in "System Management" if the AI model is correctly initiated.
For server-type machines, use 'Config' to confirm that the required AI model is active.
For more detailed instructions, refer to:
If after all checks, you still don’t see detection frames, your AI model’s accuracy might not be suitable for your current scene. Possible reasons include:
Low camera resolution causing blurry images.
Image color and brightness significantly different from typical scenarios.
Distant or small objects that AI fails to detect. For instance, with a 720p camera, objects less than 20px in size may go undetected by AI. To remedy this, we recommend adjusting the camera's shooting distance and angle.
Blurred object boundaries or objects easily confused with others can lead to inaccurate AI detection. If it's difficult for the human eye to discern object boundaries or distinguish between object types within a short period, AI may encounter similar issues.
💡 Consider AI retraining to improve accuracy.
You can label and retrain AI models with the data collected for more accurate detection fitting your specific environment.
For more detailed instructions, refer to:
If these steps don't solve the issue, please .
