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

Feature pyramids are widely exploited by both the state-of-the-art one-stageobject detectors (e.g., DSSD, RetinaNet, RefineDet) and the two-stage objectdetectors (e.g., Mask R-CNN, DetNet) to alleviate the problem arising fromscale variation across object instances. Although these object detectors withfeature pyramids achieve encouraging results, they have some limitations due tothat they only simply construct the feature pyramid according to the inherentmulti-scale, pyramidal architecture of the backbones which are actuallydesigned for object classification task. Newly, in this work, we present amethod called Multi-Level Feature Pyramid Network (MLFPN) to construct moreeffective feature pyramids for detecting objects of different scales. First, wefuse multi-level features (i.e. multiple layers) extracted by backbone as thebase feature. Second, we feed the base feature into a block of alternatingjoint Thinned U-shape Modules and Feature Fusion Modules and exploit thedecoder layers of each u-shape module as the features for detecting objects.Finally, we gather up the decoder layers with equivalent scales (sizes) todevelop a feature pyramid for object detection, in which every feature mapconsists of the layers (features) from multiple levels. To evaluate theeffectiveness of the proposed MLFPN, we design and train a powerful end-to-endone-stage object detector we call M2Det by integrating it into the architectureof SSD, which gets better detection performance than state-of-the-art one-stagedetectors. Specifically, on MS-COCO benchmark, M2Det achieves AP of 41.0 atspeed of 11.8 FPS with single-scale inference strategy and AP of 44.2 withmulti-scale inference strategy, which is the new state-of-the-art results amongone-stage detectors. The code will be made available on\url{https://github.com/qijiezhao/M2Det.

M2Det: A Single-Shot Object Detector based on Multi-Level Feature Pyramid Network