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

Weakly supervised video object localization (WSVOL) allows locating object invideos using only global video tags such as object class. State-of-art methodsrely on multiple independent stages, where initial spatio-temporal proposalsare generated using visual and motion cues, then prominent objects areidentified and refined. Localization is done by solving an optimization problemover one or more videos, and video tags are typically used for videoclustering. This requires a model per-video or per-class making for costlyinference. Moreover, localized regions are not necessary discriminant becauseof unsupervised motion methods like optical flow, or because video tags arediscarded from optimization. In this paper, we leverage the successful classactivation mapping (CAM) methods, designed for WSOL based on still images. Anew Temporal CAM (TCAM) method is introduced to train a discriminant deeplearning (DL) model to exploit spatio-temporal information in videos, using anaggregation mechanism, called CAM-Temporal Max Pooling (CAM-TMP), overconsecutive CAMs. In particular, activations of regions of interest (ROIs) arecollected from CAMs produced by a pretrained CNN classifier to build pixel-wisepseudo-labels for training the DL model. In addition, a global unsupervisedsize constraint, and local constraint such as CRF are used to yield moreaccurate CAMs. Inference over single independent frames allows parallelprocessing of a clip of frames, and real-time localization. Extensiveexperiments on two challenging YouTube-Objects datasets for unconstrainedvideos, indicate that CAM methods (trained on independent frames) can yielddecent localization accuracy. Our proposed TCAM method achieves a newstate-of-art in WSVOL accuracy, and visual results suggest that it can beadapted for subsequent tasks like visual object tracking and detection. Code ispublicly available.

TCAM: Temporal Class Activation Maps for Object Localization in Weakly-Labeled Unconstrained Videos