Papers › Exploring Enhanced Contextual Information for Video-Level Object Tracking

Exploring Enhanced Contextual Information for Video-Level Object Tracking

15 Dec 2024AAAI2025 2024 12arXiv:2412.11023archive 2025-07-28

Ben Kang, Xin Chen, Simiao Lai, Yang Liu, Yi Liu, Dong Wang

Contextual information at the video level has become increasingly crucial for visual object tracking. However, existing methods typically use only a few tokens to convey this information, which can lead to information loss and limit their ability to fully capture the context. To address this issue, we propose a new video-level visual object tracking framework called MCITrack. It leverages Mamba's hidden states to continuously record and transmit extensive contextual information throughout the video stream, resulting in more robust object tracking. The core component of MCITrack is the Contextual Information Fusion module, which consists of the mamba layer and the cross-attention layer. The mamba layer stores historical contextual information, while the cross-attention layer integrates this information into the current visual features of each backbone block. This module enhances the model's ability to capture and utilize contextual information at multiple levels through deep integration with the backbone. Experiments demonstrate that MCITrack achieves competitive performance across numerous benchmarks. For instance, it gets 76.6% AUC on LaSOT and 80.0% AO on GOT-10k, establishing a new state-of-the-art performance. Code and models are available at https://github.com/kangben258/MCITrack.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

kangben258/MCITrack officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

ObjectObject TrackingSemi-Supervised Video Object SegmentationVideo Object TrackingVisual Object TrackingVisual Tracking

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Semi-Supervised Video Object Segmentation VOT2020 MCITrack-L384 EAO 0.624 #2 of 20 Archive leaderboard report
Semi-Supervised Video Object Segmentation VOT2020 MCITrack-B224 EAO 0.619 #4 of 20 Archive leaderboard report
Visual Object Tracking GOT-10k MCITrack-L384 Average Overlap 80.0 #6 of 42 Archive leaderboard report
Visual Object Tracking GOT-10k MCITrack-L384 Success Rate 0.5 88.5 #6 of 42 Archive leaderboard report
Visual Object Tracking GOT-10k MCITrack-L384 Success Rate 0.75 80.2 #6 of 42 Archive leaderboard report
Visual Object Tracking GOT-10k MCITrack-B224 Average Overlap 77.9 #11 of 42 Archive leaderboard report
Visual Object Tracking GOT-10k MCITrack-B224 Success Rate 0.5 88.2 #11 of 42 Archive leaderboard report
Visual Object Tracking GOT-10k MCITrack-B224 Success Rate 0.75 76.8 #11 of 42 Archive leaderboard report
Visual Object Tracking LaSOT MCITrack-L384 AUC 76.6 #3 of 46 Archive leaderboard report
Visual Object Tracking LaSOT MCITrack-L384 Normalized Precision 86.1 #3 of 46 Archive leaderboard report
Visual Object Tracking LaSOT MCITrack-L384 Precision 85.0 #3 of 46 Archive leaderboard report
Visual Object Tracking LaSOT MCITrack-B224 AUC 75.3 #5 of 46 Archive leaderboard report
Visual Object Tracking LaSOT MCITrack-B224 Normalized Precision 85.6 #5 of 46 Archive leaderboard report
Visual Object Tracking LaSOT MCITrack-B224 Precision 83.3 #5 of 46 Archive leaderboard report
Visual Object Tracking LaSOT-ext MCITrack-L384 AUC 55.7 #6 of 18 Archive leaderboard report
Visual Object Tracking LaSOT-ext MCITrack-L384 Normalized Precision 66.5 #6 of 18 Archive leaderboard report
Visual Object Tracking LaSOT-ext MCITrack-L384 Precision 62.9 #6 of 18 Archive leaderboard report
Visual Object Tracking LaSOT-ext MCITrack-B224 AUC 54.6 #8 of 18 Archive leaderboard report
Visual Object Tracking LaSOT-ext MCITrack-B224 Normalized Precision 65.7 #8 of 18 Archive leaderboard report
Visual Object Tracking LaSOT-ext MCITrack-B224 Precision 62.1 #8 of 18 Archive leaderboard report
Visual Object Tracking TNL2K MCITrack-L384 AUC 65.3 #1 of 16 Archive leaderboard report
Visual Object Tracking TNL2K MCITrack-B224 AUC 62.9 #4 of 16 Archive leaderboard report
Visual Object Tracking TrackingNet MCITrack-L384 Accuracy 87.9 #1 of 40 Archive leaderboard report
Visual Object Tracking TrackingNet MCITrack-L384 Normalized Precision 92.1 #1 of 40 Archive leaderboard report
Visual Object Tracking TrackingNet MCITrack-L384 Precision 89.2 #1 of 40 Archive leaderboard report
Visual Object Tracking TrackingNet MCITrack-B224 Accuracy 86.3 #4 of 40 Archive leaderboard report
Visual Object Tracking TrackingNet MCITrack-B224 Normalized Precision 90.9 #4 of 40 Archive leaderboard report
Visual Object Tracking TrackingNet MCITrack-B224 Precision 86.1 #4 of 40 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AOMamba

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections