A framework segments panoramic video into sub-images for detection, modifies multi-object tracking for boundary continuity, and applies it to vehicle overtaking detection in real cycling videos, reporting gains in precision and an F-score of 0.82.
154 Existing object detection and MOT algorithms are trained on video from cameras with155 a limited field of view (FOV)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Multiple Object Detection and Tracking in Panoramic Videos for Cycling Safety Analysis
A framework segments panoramic video into sub-images for detection, modifies multi-object tracking for boundary continuity, and applies it to vehicle overtaking detection in real cycling videos, reporting gains in precision and an F-score of 0.82.