Video Task (Beta)
Last updated
Last updated
Video detection tasks are one-time AI detections designed to simulate real-time detection scenarios.
Through video tasks, you can adjust various AI and event parameters by testing detection results repeatedly, verifying your AI detection scenario plans.
Navigate to 'Resource', switch to 'Video', view current videos, and click '+ Video' to add a new video.
Video Name: Name of the video
Location: Location related to the video event
Upload Video: Click to upload your video. Supported formats include MP4, AVI, MPEG, FLV, with file sizes under 100MB.
Once the video is ready, you can initiate one or more detection tasks.
In the Task page, click '+ Task' at the top right corner and select 'Video Task' for single-camera detection tasks, leading to the new task creation page.
Step One: Set Task Information Once all information on the page is set, click 'Next' to proceed.
Task Name: Define the name for this task.
Organization and Location: Select the organization and location of the current task's camera.
Compute Target: Choose the computing machine for detection, noting that only x86 machines support video detection tasks.
Video: Select the video to be detected (previously uploaded).
Step Two: Choose AI Model Detection Scenario Click 'Next' after selection to proceed.
The platform offers dozens of AI detection scenarios including electronic fencing, PPE detection, construction operations, smoke detection, stacking detection, people and vehicle counting, and more.
Note: Different versions offer different AI detection scenarios and initial parameter settings.
Click 'Next' to edit detection events. Once settings are complete, click 'Create' to add the task.
Check Events: Select AI detection events 'Event'.
Notification: Decide if notifications should be sent 'Notification'.
Notification Recipients: Set up users to be notified when events occur.
Once all settings are finalized, the newly created video task will be visible in the Task List.
Switch to 'Video' in the Task List to view all current video detection tasks.
Click on a task to set up video task settings.
Video tasks are one-time detections; you need to set the following before detection:
Event Zone: Reference + Set Event Zone
AI Threshold: Reference + Set AI Confidence Score
Event Threshold: Reference + Set Event Parameters
After setting up, click 'Start Detection', and the system will begin detection.
Quick Tip: You can repeatedly perform video detections to adjust suitable AI/event parameters.
The system requires some time to perform AI detection, varying with video length.
After detection, you will see the event detection results on the following screen. Click 'Settings', switch to 'Video Detection Result', and you can view the complete detection video.
Note: Video tasks primarily simulate real-time detection scenarios.
Video detection results are based on real-time streaming analysis, with processing optimizations that may skip frames to enhance performance, potentially leading to a final result video length different from the original video.
If you need results for every frame, perform Data Slice Prediction on DataVerse for detailed frame-by-frame analysis.