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/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

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

Zendy Tools

Zendy Open

High computational power and significant time are usually needed to train adeep learning based tracker on large datasets. Depending on many factors,training might not always be an option. In this paper, we propose a frameworkwith two ideas on Siamese-based trackers. (i) Extending number of templates ina way that removes the need to retrain the network and (ii) a lightweighttemporal network with a novel architecture focusing on both local and globalinformation that can be used independently from trackers. Most Siamese-basedtrackers only rely on the first frame as the ground truth for objects andstruggle when the target's appearance changes significantly in subsequentframes in presence of similar distractors. Some trackers use multiple templateswhich mostly rely on constant thresholds to update, or they replace thosetemplates that have low similarity scores only with more similar ones. Unlikeprevious works, we use adaptive thresholds that update the bag with similartemplates as well as those templates which are slightly diverse. Adaptivethresholds also cause an overall improvement over constant ones. In addition,mixing feature maps obtained by each template in the last stage of networksremoves the need to retrain trackers. Our proposed lightweight temporalnetwork, CombiNet, learns the path history of different objects using onlyobject coordinates and predicts target's potential location in the next frame.It is tracker independent and applying it on new trackers does not need furthertraining. By implementing these ideas, trackers' performance improved on alldatasets tested on, including LaSOT, LaSOT extension, TrackingNet, OTB100,OTB50, UAV123 and UAV20L. Experiments indicate the proposed framework workswell with both convolutional and transformer-based trackers. The officialpython code for this paper will be publicly available upon publication.

Improving Siamese Based Trackers with Light or No Training through Multiple Templates and Temporal Network