WikiChurches: A Fine-Grained Dataset of Architectural Styles with Real-World Challenges (2021-08-16T00:00:00.000000Z)

TL;DR

A novel dataset for architectural style classification, consisting of 9,485 images of church buildings, provides 631 bounding box annotations of characteristic visual features for 139 churches from four major categories, to be useful for research on fine-grained classification.

Abstract

We introduce a novel dataset for architectural style classification, consisting of 9,485 images of church buildings. Both images and style labels were sourced from Wikipedia. The dataset can serve as a benchmark for various research fields, as it combines numerous real-world challenges: fine-grained distinctions between classes based on subtle visual features, a comparatively small sample size, a highly imbalanced class distribution, a high variance of viewpoints, and a hierarchical organization of labels, where only some images are labeled at the most precise level. In addition, we provide 631 bounding box annotations of characteristic visual features for 139 churches from four major categories. These annotations can, for example, be useful for research on fine-grained classification, where additional expert knowledge about distinctive object parts is often available. Images and annotations are available at: https://doi.org/10.5281/zenodo.5166987

Authors

Joachim Denzler

7 papers

Björn Barz

4 papers

TL;DR

Abstract

Authors

References35 items

VIPriors 3: Visual Inductive Priors for Data-Efficient Deep Learning Challenges

The Semi-Supervised iNaturalist Challenge at the FGVC8 Workshop

The Herbarium 2021 Half-Earth Challenge Dataset

ImageNet-21K Pretraining for the Masses

MonuMAI: Dataset, deep learning pipeline and citizen science based app for monumental heritage taxonomy and classification

Making Every Label Count: Handling Semantic Imprecision by Integrating Domain Knowledge

End-to-End Object Detection with Transformers

An Evaluation of Deep Learning Methods for Small Object Detection

Classification-Specific Parts for Improving Fine-Grained Visual Categorization

Deep Learning on Small Datasets without Pre-Training using Cosine Loss

Integrating domain knowledge: using hierarchies to improve deep classifiers

A System for Massively Parallel Hyperparameter Tuning

Places: A 10 Million Image Database for Scene Recognition

Region-Based Image Retrieval Revisited

Classification of Architectural Heritage Images Using Deep Learning Techniques

The iNaturalist Species Classification and Detection Dataset

SGDR: Stochastic Gradient Descent with Warm Restarts

Deep Residual Learning for Image Recognition

Rethinking the Inception Architecture for Computer Vision

Interactive Image Retrieval for Biodiversity Research

Wikidata

Large-Scale Object Classification Using Label Relation Graphs

Architectural Style Classification Using Multinomial Latent Logistic Regression

ImageNet Large Scale Visual Recognition Challenge

Nonparametric Part Transfer for Fine-Grained Recognition

3D Object Representations for Fine-Grained Categorization

The Caltech-UCSD Birds-200-2011 Dataset

ImageNet: A large-scale hierarchical image database

Semantic Hierarchies for Visual Object Recognition

The Open Images Dataset V4: Uniﬁed image classiﬁcation, object detection, and visual relationship detection at scale

SPARQL query language for RDF

Hyper-parameters used for training the three baseline models on WikiChurches-6. Pre-Training Balancing Resolution Epochs Batch Size Learning Rate Weight Decay

WikiChurches-6 is a subset of WikiChurches-14 restricted to the 6 largest classes comprising more than 200 images. These few classes account for 89% of all images in WikiChurches

intended for studying hierarchical classiﬁcation

WikiChurches-4 is a subset of WikiChurches-6 limited to the four classes for which we provide bounding box annotations of characteristic visual features (see Sections 2.4 and 3.6)

Field of Study

Journal Information

Name

Volume