English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Discriminative representation is crucial for the association step inmulti-object tracking. Recent work mainly utilizes features in single orneighboring frames for constructing metric loss and empowering networks toextract representation of targets. Although this strategy is effective, itfails to fully exploit the information contained in a whole trajectory. To thisend, we propose a strategy, namely multi-view trajectory contrastive learning,in which each trajectory is represented as a center vector. By maintaining allthe vectors in a dynamically updated memory bank, a trajectory-levelcontrastive loss is devised to explore the inter-frame information in the wholetrajectories. Besides, in this strategy, each target is represented as multipleadaptively selected keypoints rather than a pre-defined anchor or center. Thisdesign allows the network to generate richer representation from multiple viewsof the same target, which can better characterize occluded objects.Additionally, in the inference stage, a similarity-guided feature fusionstrategy is developed for further boosting the quality of the trajectoryrepresentation. Extensive experiments have been conducted on MOTChallenge toverify the effectiveness of the proposed techniques. The experimental resultsindicate that our method has surpassed preceding trackers and established newstate-of-the-art performance.

Towards Discriminative Representation: Multi-view Trajectory Contrastive Learning for Online Multi-object Tracking