Why is this needed:
Person Re-ID based on Cloud-edge collaborative architecture is a new example feature that Sedna will soon support, which can continuously recognize, track and search the target person provided by the user in the source video, and push the video containing the search results to the streaming media server. Since each service module flexibly supports the AI model, it is necessary to provide users with the actual performance test of THE AI algorithm to provide a reference for the actual search effect of the system.
What would you like to be added/modified:
This function is designed to help users to know the performance of each algorithm module of Re-ID, so we should make it :
Supports the existing capabilities and runtime of Sedna and integrate into the Re-ID features;
Use open-source data sets to create test reports automately, need to define metrics, related interface, preprocessing, and etc;
Why is this needed: Person Re-ID based on Cloud-edge collaborative architecture is a new example feature that Sedna will soon support, which can continuously recognize, track and search the target person provided by the user in the source video, and push the video containing the search results to the streaming media server. Since each service module flexibly supports the AI model, it is necessary to provide users with the actual performance test of THE AI algorithm to provide a reference for the actual search effect of the system.
What would you like to be added/modified: This function is designed to help users to know the performance of each algorithm module of Re-ID, so we should make it :
Other information: Recommended Skills: Pytorch, Python, K8s Kubeedge: https://github.com/kubeedge/kubeedge; Sedna: https://github.com/kubeedge/sedna; Sedna Re-ID: https://github.com/vcozzolino/sedna/tree/feature-reid; Sedna Re-ID tutorial: https://github.com/vcozzolino/sedna/blob/feature-reid/examples/multiedgetracking/tutorial/README.md; KubeEdge SIG AI: https://github.com/kubeedge/community/tree/master/sig-ai