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

Few-shot object detection (FSOD) aims to achieve object detection only usinga few novel class training data. Most of the existing methods usually adopt atransfer-learning strategy to construct the novel class distribution bytransferring the base class knowledge. However, this direct way easily resultsin confusion between the novel class and other similar categories in thedecision space. To address the problem, we propose generating local reversesamples (LRSamples) in Prototype Reference Frames to adaptively adjust thecenter position and boundary range of the novel class distribution to learnmore discriminative novel class samples for FSOD. Firstly, we propose a CenterCalibration Variance Augmentation (CCVA) module, which contains the selectionrule of LRSamples, the generator of LRSamples, and augmentation on thecalibrated distribution centers. Specifically, we design an intra-class featureconverter (IFC) as the generator of CCVA to learn the selecting rule. Bytransferring the knowledge of IFC from the base training to fine-tuning, theIFC generates plentiful novel samples to calibrate the novel classdistribution. Moreover, we propose a Feature Density Boundary Optimization(FDBO) module to adaptively adjust the importance of samples depending on theirdistance from the decision boundary. It can emphasize the importance of thehigh-density area of the similar class (closer decision boundary area) andreduce the weight of the low-density area of the similar class (fartherdecision boundary area), thus optimizing a clearer decision boundary for eachcategory. We conduct extensive experiments to demonstrate the effectiveness ofour proposed method. Our method achieves consistent improvement on the PascalVOC and MS COCO datasets based on DeFRCN and MFDC baselines.

GRSDet: Learning to Generate Local Reverse Samples for Few-shot Object Detection