Motion propagation detection association for multi-target tracking in wide area aerial surveillance

Chen, Bor-Jeng; Medioni, Gérard

doi:10.1109/avss.2015.7301766

Cited by 21 publications

(21 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using LCT further improves more than 19% in recall from move-then-stop situations. Methods that rely on only motion detections [9,38,21,7] fail in these cases. This sequence has higher ground sampling distance and lower contrast than the WPFAB 2009 sequence.…”

Section: Results On Rochester Sequencementioning

confidence: 99%

“…We compare our methods with state-of-the-art trackers [9,38,6,21,7] on two WAMI sequences. In order to get a fair comparison, we use the same motion detection result as input for all trackers.…”

Section: Setupmentioning

confidence: 99%

“…We obtain executables of [9] and [6] from the authors of [6]. The programs of [38,7] are provided by their authors. We do not change parameters for the…”

Section: Setupmentioning

confidence: 99%

“…Our quantitative evaluation is based on commonly used metrics which are also adopted in [6,7], including recall : number of true positive detection/number of ground truth detection; precision: number of true positive detection/number of detection in tracks; false positive per frame (FP/F); false positive per ground truth detection(FP/GT); multiple object detection accuracy (MODA); number of swaps (id-switches) per track (S/T); number of breaks per track (B/T); multiple object tracking accuracy (MOTA). The definition of MODA and MOTA can be found in [47].…”

Section: Evaluation Metricsmentioning

confidence: 99%

“…With current sensor and storage technologies, WAMI is typically captured in large format (tens ∼ hundreds of megapixels), low frame rate (1 ∼ 2 Hz), grayscale, and with ground sampling distance from 0.2 ∼ 0.5 meter/pixel. Because of these unique characteristics and its wide applications, it has gained attention in the computer vision field in recent years [1,2,3,4,5,6,7]. Tracking many vehicles in WAMI [6,7,8,9,10] is an essential component since it is needed for higher level activity analysis and scene understanding.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Exploring Local Context for Multi-target Tracking in Wide Area Aerial Surveillance

Chen

Medioni

2017

2017 IEEE Winter Conference on Applications of Computer Vision (WACV)

Self Cite

View full text Add to dashboard Cite

Abstract. Tracking many vehicles in wide coverage aerial imagery is crucial for understanding events in a large field of view. Most approaches aim to associate detections from frame differencing into tracks. However, slow or stopped vehicles result in long-term missing detections and further cause tracking discontinuities. Relying merely on appearance clue to recover missing detections is difficult as targets are extremely small and in grayscale. In this paper, we address the limitations of detection association methods by coupling it with a local context tracker (LCT), which does not rely on motion detections. On one hand, our LCT learns neighboring spatial relation and tracks each target in consecutive frames using graph optimization. It takes the advantage of context constraints to avoid drifting to nearby targets. We generate hypotheses from sparse and dense flow efficiently to keep solutions tractable. On the other hand, we use detection association strategy to extract short tracks in batch processing. We explicitly handle merged detections by generating additional hypotheses from them. Our evaluation on wide area aerial imagery sequences shows significant improvement over state-of-the-art methods.

show abstract

Section: Results On Rochester Sequencementioning

confidence: 99%

Section: Setupmentioning

confidence: 99%

“…We obtain executables of [9] and [6] from the authors of [6]. The programs of [38,7] are provided by their authors. We do not change parameters for the…”

Section: Setupmentioning

confidence: 99%