# Towards Sequence-Level Training for Visual Tracking

Source: [Towards Sequence-Level Training for Visual Tracking](<https://www.sota2.com/research/paper/68fa6dab1e209131cea3b5e9>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T03:45:27.317Z

Authors: Minji Kim, Seungkwan Lee, Jungseul Ok, Bohyung Han, Minsu Cho

Published: 2022-08-11T13:15:36.000Z

## Abstract

Despite the extensive adoption of machine learning on the task of visual object tracking, recent learning-based approaches have largely overlooked the fact that visual tracking is a sequence-level task in its nature; they rely heavily on frame-level training, which inevitably induces inconsistency between training and testing in terms of both data distributions and task objectives. This work introduces a sequence-level training strategy for visual tracking based on reinforcement learning and discusses how a sequence-level design of data sampling, learning objectives, and data augmentation can improve the accuracy and robustness of tracking algorithms. Our experiments on standard benchmarks including LaSOT, TrackingNet, and GOT-10k demonstrate that four representative tracking models, SiamRPN++, SiamAttn, TransT, and TrDiMP, consistently improve by incorporating the proposed methods in training without modifying architectures.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Visual Object Tracking on GOT-10k Restricted Protocol (test)](<https://www.sota2.com/research/sota/visual-object-tracking-on-got-10k-restricted-protocol-test.md>): Visual Object Tracking; GOT-10k Restricted Protocol (test); AO: 67.5.

- [Visual Object Tracking on LaSOT 2019 (test)](<https://www.sota2.com/research/sota/visual-object-tracking-on-lasot-2019-test.md>): Visual Object Tracking; LaSOT 2019 (test); AUC: 66.8.

- [Visual Object Tracking on TrackingNet (test)](<https://www.sota2.com/research/sota/visual-object-tracking-on-trackingnet-test.md>): Visual Object Tracking; TrackingNet (test); Normalized Precision (Pnorm): 87.5.

- [Visual Object Tracking on GOT-10k (test)](<https://www.sota2.com/research/sota/visual-object-tracking-on-got-10k-test.md>): Visual Object Tracking; GOT-10k (test); Average Overlap: 72.5.

- [Visual Object Tracking on LaSOT (test)](<https://www.sota2.com/research/sota/visual-object-tracking-on-lasot-test.md>): Visual Object Tracking; LaSOT (test); AUC: 66.8.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2208.05810>)

[PDF](<https://arxiv.org/pdf/2208.05810.pdf>)

[Implementation](<https://github.com/byminji/SLTtrack>)