In the context of multi-objective multi-class action detection, does the algorithm employ a multi-object tracking (MOT) method to associate the same target across video frames, forming tracklets? This would ensure consistent spatiotemporal feature extraction for each individual target throughout the sequence.
In the context of multi-objective multi-class action detection, does the algorithm employ a multi-object tracking (MOT) method to associate the same target across video frames, forming tracklets? This would ensure consistent spatiotemporal feature extraction for each individual target throughout the sequence.