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

The counting task, which plays a fundamental role in numerous applications(e.g., crowd counting, traffic statistics), aims to predict the number ofobjects with various densities. Existing object counting tasks are designed fora single object class. However, it is inevitable to encounter newly coming datawith new classes in our real world. We name this scenario as \textit{evolvingobject counting}. In this paper, we build the first evolving object countingdataset and propose a unified object counting network as the first attempt toaddress this task. The proposed model consists of two key components: aclass-agnostic mask module and a class-incremental module. The class-agnosticmask module learns generic object occupation prior via predicting aclass-agnostic binary mask (e.g., 1 denotes there exists an object at theconsidering position in an image and 0 otherwise). The class-incremental moduleis used to handle new coming classes and provides discriminative class guidancefor density map prediction. The combined outputs of class-agnostic mask moduleand image feature extractor are used to predict the final density map. When newclasses come, we first add new neural nodes into the last regression andclassification layers of class-incremental module. Then, instead of retrainingthe model from scratch, we utilize knowledge distillation to help the modelremember what have already learned about previous object classes. We alsoemploy a support sample bank to store a small number of typical trainingsamples of each class, which are used to prevent the model from forgetting keyinformation of old data. With this design, our model can efficiently andeffectively adapt to new coming classes while keeping good performance onalready seen data without large-scale retraining. Extensive experiments on thecollected dataset demonstrate the favorable performance.

A Unified Object Counting Network with Object Occupation Prior