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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Effective search for engineering diagram images in larger collections is challenging because most existing feature extraction models are pre-trained on natural image data rather than diagrams. Surprisingly, we observe through experiments that even in-domain training with standard unsupervised representation learning techniques leads to poor results. We argue that, because of their structured nature, diagram images require more specially-tailored learning objectives. We propose a new method for unsupervised adaptation of out-of-domain feature extractors that asks the model to reason about spatial context. Specifically, we fine-tune a pre-trained image encoder by requiring it to correctly predict the relative orientation between pairs of nearby image regions. Experiments on the recently released Ikea Diagram Dataset show that our proposed method leads to substantial improvements on a downstream search task, more than doubling recall for certain query categories in the dataset.

Domain Adaptation via Context Prediction for Engineering Diagram Search