Evaluate Predicted Identity

[hobbitsyfeet]

To evaluate the predicted identity, we can't evaluate the model's performance with identity without assigning one.

We can do this in 2 ways. K Nearest neighbour where the nearest rectangle is the chosen one.

• KNN centroid
• KNN box corners
• KNN box corners with centroid

Kalman Filter with KNN

• KNN with Kalman predicted centroid
• KNN with Kalman predicted box corners
• KNN with Kalman predicted corners with centroid

[hobbitsyfeet]

Completed somewhere around here https://github.com/hobbitsyfeet/PeopleTracker/commit/627e8eee2232c2630ba3f384799ce68be3f9544e : We used and modified: https://www.pyimagesearch.com/2018/07/23/simple-object-tracking-with-opencv/