Hard to Track Objects with Irregular Motions and Similar Appearances? Make It Easier by Buffering the Matching Space

Motivation

• Irregular motions and indistinguishable appearances challenge conventional MOT methods, causing performance drops on datasets such as DanceTrack.Methods that perform well on MOT17 may fail when objects move irregularly and look similar.Source: §I · p. 1

Buffered IoU

• BIoU expands the matching space without changing bounding-box centers, scale ratios, or shapes.The buffer is proportional to the original bounding box.Source: Figure 2 · p. 2, §I · p. 2
• BIoU directly matches identical objects whose detections and tracks do not overlap in adjacent frames.This addresses mismatches caused by fast movement or inaccurate motion estimation.Source: Figure 3 · p. 2, §I · p. 2

Cascaded Matching

• Cascaded matching first uses a small buffer for alive tracks and detections, then a large buffer for unmatched pairs.The cascade reduces the risk of overexpanding the matching space.Source: §I · p. 2

Results

• 2.6 to 7.2 HOTA points: C-BIoU improves tracking across MOT datasets focused on irregular motions and indistinguishable appearances.The tracker was also the dominant component of second-place solutions in the SoccerNet MOT and MOTComplex DanceTrack challenges.Source: §I · p. 2

Appearance Consistency

• Appearance consistency is useful for cross-frame association, but it becomes unreliable when tracking targets have similar appearances.DanceTrack, SoccerNet, and GMOT-40 expose this challenge.Source: A. Appearance Consistency and… · p. 2

Geometric Consistency

• IoU is commonly used to measure geometric consistency between detections and track predictions.Motion estimation commonly uses Kalman filters, while other approaches use neural networks or Bayesian filters.Source: A. Appearance Consistency and… · p. 2
• Irregular motions by soccer players and dancers can cause motion-estimation models to fail.Non-stationary cameras add further difficulty because image registration is time-consuming and its accuracy is not guaranteed.Source: A. Appearance Consistency and… · p. 3

Geometric Consistency Measurement

• IoU assigns zero similarity to all non-overlapping boxes, which can break initialization and ongoing tracking under irregular motion.BIoU expands the matching space to measure geometric consistency despite fast motion and estimation bias.Source: B. Geometric Consistency Measurement · p. 3

Cascaded Matching

• C-BIoU first matches alive tracks and detections with a small buffer, then matches remaining pairs with a large buffer.This ordering follows the principle of matching confident samples before ambiguous ones.Source: C. Cascaded Matching · p. 3

Tracker Design

• The tracker inherits part of SORT's track management and uses C-BIoU for geometric consistency measurement.Its design targets objects with indistinguishable appearances and irregular motions.Source: §III · p. 3

Tracking Pipeline

• The pipeline applies an off-the-shelf detector to generate bounding boxes, which C-BIoU then uses to produce tracking results.Separating detection from tracking allows the tracker to operate on arbitrary detections.Source: A. Tracking Pipeline · p. 3

Buffered IoU

• BIoU buffers are proportional to detections and tracks while preserving their centers, scale ratios, and shapes.Two buffer scales are selected by grid search for cascaded matching.Source: B. Buffered IoU · p. 3

Simple Motion Estimation

• C-BIoU averages motions from recent matched detections instead of using a Kalman filter for state estimation.The default number of recent frames is 2 ≤ n ≤ 5.Source: C. Simple Motion Estimation · p. 3
• The simple motion-estimation approach increases tracking speed by reducing computation associated with Kalman-filter calculations.The buffered track state is used for data association at later frames.Source: C. Simple Motion Estimation · p. 3

• New tracks initially use their recorded bounding boxes as states because motion cannot yet be estimated reliably.Source: D. Track Management · pp. 3–4
• Tracks are initialized from unmatched detections and terminated after remaining unmatched for a specified max age.Source: D. Track Management · p. 3, Figure 4 · p. 4
• The first association matches alive tracks and detections using BIoU with the small buffer scale b1 and linear assignment.Source: D. Track Management · p. 4, Figure 4 · p. 4
• A second association applies BIoU with the large buffer scale b2 to unmatched tracks and detections affected by large irregular motions.Source: D. Track Management · p. 4, Figure 4 · p. 4
• The prototype can be combined with other MOT modules to construct a more powerful tracking framework.Source: D. Track Management · p. 4

Experimental Design

• The experiments compare C-BIoU with state-of-the-art methods on four MOT datasets using official or prior-study results where applicable.Source: §IV · p. 4, A. Dataset and Evaluation Metrics · p. 4

Ablation Studies

• Ablation studies examine BIoU, cascaded matching, motion modeling, buffer scales, detection noise, and corresponding limitations.Source: §IV · p. 4

Datasets

• Four datasets cover conventional MOT17 scenes and irregular-motion, similar-appearance scenarios in DanceTrack, SoccerNet, and GMOT-40.Source: A. Dataset and Evaluation Metrics · p. 4

Evaluation Metrics

• HOTA, DetA, AssA, MOTA, and IDF1 evaluate tracking from detection, association, overall MOT, and identity perspectives.Source: A. Dataset and Evaluation Metrics · p. 4
• Tracking inference speed is reported in FPS without the detection component, using an Intel Xeon Silver 4216 CPU.Source: Table II · p. 5

• On MOT17, C-BIoU achieves a similar HOTA score to other methods, while DanceTrack exposes larger tracking challenges.Source: 1) Comparisons Using Estimated… · p. 5
• 4.9 HOTA points separate C-BIoU from OC-SORT on DanceTrack, establishing a new state of the art in that comparison.Source: 1) Comparisons Using Estimated… · p. 5
• Almost 2× the tracking inference speed of OC-SORT is achieved on MOT17 and DanceTrack without the detection component.Source: 1) Comparisons Using Estimated… · p. 5

• 2.6 to 7.2 HOTA points of improvement are reported over other methods across DanceTrack, SoccerNet, and GMOT-40 with oracle detections.Source: 2) Comparisons Using Oracle Detections · p. 5
• Oracle detections let the experiments focus on tracking performance rather than detection quality.Source: 2) Comparisons Using Oracle Detections · pp. 5–6
• Even with oracle detections, the reported tracking results are still imperfect.Source: 2) Comparisons Using Oracle Detections · p. 5
• C-BIoU generates fewer tracking errors than SORT and OC-SORT in examples from DanceTrack and SoccerNet.Source: Figure 5 · p. 6

Effect of the BIoU

• BIoU achieves higher HOTA than IoU, GIoU, and DIoU within the same tracker framework.The comparison removes cascaded matching and motion estimation to isolate geometric matching.Source: 1) Effect of Each Module in the C-BIoU… · p. 6

Effect of Cascaded Matching

• A HOTA gain of 1.2 results from integrating cascaded matching and BIoU on SoccerNet.The non-stationary camera produces extremely fast object motion, making cascade matching more robust in this setting.Source: 1) Effect of Each Module in the C-BIoU… · p. 6

Effect of Motion Estimation

• Motion estimation improves HOTA when combined with BIoU, even when implemented by averaging previous motions.BIoU compensates the matching space for incorrect motion estimation.Source: 1) Effect of Each Module in the C-BIoU… · p. 6

Effect of Buffer Scales

• The [0.3, 0.4] buffer-scale combination gives the maximum HOTA score on the DanceTrack validation set.The study evaluates 21 combinations with b1 < b2, where both scales range from 0.1 to 0.7.Source: 2) Effect of Buffer Scales in the… · p. 6, Figure 6 · p. 7

Effect of Detection Noise

• The authors evaluate detection-quality effects by injecting matched-ratio false negatives and false positives into oracle DanceTrack detections.Detections are first removed to create false negatives, then added at non-target locations to create false positives.Source: 3) Effect of the Detection Noise · p. 7, Table V · p. 7
• At a 40% noise ratio, C-BIoU can perform worse than the normal IoU tracker.High-ratio noisy detections leave tracks and detections non-overlapping for longer intervals and make buffered matching more aggressive toward false positives.Source: 3) Effect of the Detection Noise · p. 7
• At noise ratios no higher than 20%, C-BIoU maintains the best performance.The buffer helps match objects whose track and detection briefly fail to overlap.Source: 3) Effect of the Detection Noise · p. 7

Conclusion

• C-BIoU expands the matching space to match identical non-overlapping detections and tracks and compensate for motion-estimation bias.These two mechanisms are presented as the basis for mitigating irregular-motion effects.Source: §V · pp. 7–8
• C-BIoU outperforms most existing methods on MOT datasets featuring indistinguishable appearances and irregular motions.The authors describe the results as suggesting generalizability within this application setting.Source: §V · p. 7

Limitation

• C-BIoU may not be robust to extremely noisy detections.The conclusion also notes simplicity, fast speed, and good tracking performance as reasons it remains applicable to real-world settings.Source: §V · p. 8