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 Plus annual

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

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Drones are employed in a growing number of visual recognition applications. Arecent development in cell tower inspection is drone-based asset surveillance,where the autonomous flight of a drone is guided by localizing objects ofinterest in successive aerial images. In this paper, we propose a method totrain deep weakly-supervised object localization (WSOL) models based only onimage-class labels to locate object with high confidence. To train ourlocalizer, pseudo labels are efficiently harvested from a self-supervisedvision transformers (SSTs). However, since SSTs decompose the scene intomultiple maps containing various object parts, and do not rely on any explicitsupervisory signal, they cannot distinguish between the object of interest andother objects, as required WSOL. To address this issue, we propose leveragingthe multiple maps generated by the different transformer heads to acquirepseudo-labels for training a deep WSOL model. In particular, a newDiscriminative Proposals Sampling (DiPS) method is introduced that relies on aCNN classifier to identify discriminative regions. Then, foreground andbackground pixels are sampled from these regions in order to train a WSOL modelfor generating activation maps that can accurately localize objects belongingto a specific class. Empirical results on the challenging TelDrone datasetindicate that our proposed approach can outperform state-of-art methods over awide range of threshold values over produced maps. We also computed results onCUB dataset, showing that our method can be adapted for other tasks.

Discriminative Sampling of Proposals in Self-Supervised Transformers for Weakly Supervised Object Localization