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

It is challenging to train a robust object detector under the supervisedlearning setting when the annotated data are scarce. Thus, previous approachestackling this problem are in two categories: semi-supervised learning modelsthat interpolate labeled data from unlabeled data, and self-supervised learningapproaches that exploit signals within unlabeled data via pretext tasks. Toseamlessly integrate and enhance existing supervised object detection methods,in this work, we focus on addressing the data scarcity problem from afundamental viewpoint without changing the supervised learning paradigm. Wepropose a new offline data augmentation method for object detection, whichsemantically interpolates the training data with novel views. Specifically, ournew system generates controllable views of training images based ondifferentiable neural rendering, together with corresponding bounding boxannotations which involve no human intervention. Firstly, we extract andproject pixel-aligned image features into point clouds while estimating depthmaps. We then re-project them with a target camera pose and render a novel-view2d image. Objects in the form of keypoints are marked in point clouds torecover annotations in new views. Our new method is fully compatible withonline data augmentation methods, such as affine transform, image mixup, etc.Extensive experiments show that our method, as a cost-free tool to enrichimages and labels, can significantly boost the performance of object detectionsystems with scarce training data. Code is available at\url{https://github.com/Guanghan/DANR}.

Data Augmentation for Object Detection via Differentiable Neural Rendering