The ArtBench Dataset: Benchmarking Generative Models with Artworks

Liao, Peiyuan; Li, Xiuyu; Liu, Xihui; Keutzer, Kurt

doi:10.48550/arxiv.2206.11404

Cited by 3 publications

(3 citation statements)

References 27 publications

(57 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(16)(17)(18). V adv is initialized by a clean image V sampled from the ArtBench dataset (Liao et al, 2022). To ensure human imperceptibility, we consider two different attack types to constrain the optimization of V adv .…”

Section: Basic Setupsmentioning

confidence: 99%

Genetic characteristics of Jiaji Duck by whole genome re-sequencing

Feng

Lin

et al. 2020

PLoS ONE

View full text Add to dashboard Cite

Jiaji Duck (JJ) is a Muscovy duck species that possesses many superior characteristics, and it has become an important genetic resource in China. However, to date, its genetic characteristics and genetic relationship with other duck breeds have not been explored yet, which greatly limits the utilization of JJ. In the present study, we investigated the genome sequences of 15 individual ducks representing five different duck populations, including JJ, French Muscovy duck (FF), mallard (YD), hong duck (HD) and Beijing duck (BD). Moreover, we investigated the characteristics of JJ-specific single nucleotide polymorphisms (SNPs) and compared the genome sequences of JJ vs. YD and JJ vs. BD using integrated strategies, including mutation detection, selective screening, and Gene Ontology (GO) analysis. More than 40 Gb of clean data were obtained for each population (mean coverage of 13.46 Gb per individual). A total number of 22,481,367 SNPs and 4,156,829 small insertion-deletions (Indels) were identified for the five duck populations, which could be used as molecular markers in breeding and utilization of JJ. Moreover, we identified 1,447,932 JJ-specific SNPs, and found that genes covering at least one JJ-specific SNP mainly involved in protein phosphorylation and dephosphorylation, as well as DNA modification. Phylogenetic tree and principal components analysis (PCA) revealed that the genetic relationship of JJ was closest to FF, while it was farthest to BD. A total of 120 and 111 genes were identified as positive selection genes for JJ vs. BD and JJ vs. YD, respectively. GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses showed that the positive selection genes for JJ vs. BD ducks mainly involved in pigmentation, muscle contraction and stretch, gland secretion, and immunology, while the positive selection genes obtained from JJ vs. YD ducks mainly involved in embryo development, muscle contraction and stretch, and gland secretion. Taken together, our findings enabled us to better understand the characteristics of JJ and provided a molecular basis for the breeding and hybrid utilization of JJ in the future.

show abstract

Section: Basic Setupsmentioning

confidence: 99%

Genetic characteristics of Jiaji Duck by whole genome re-sequencing

Feng

Lin

et al. 2020

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…3 We also excluded data sets that are exclusively applicable to other research areas like aesthetic quality assessment [2], sentiment analysis [54,88], or correspondence matching [35,70]. While formal attributes at the image-level are contained in a large number of data sets, enabling the classification of artists, materials, or creation dates, among others [5,41,46,50,52,55,75,76,85,90], content-based tags are less frequent. This is due to the fact that labels referring to the image phenomena actually shown must be determined by manual annotation, driven either by crowdsourcing approaches [4] or singular institutional efforts [16,29,59].…”

Section: Related Workmentioning

confidence: 99%

Poses of People in Art: A Data Set for Human Pose Estimation in Digital Art History

Schneider,

Vollmer,

Huber

2023

View full text Add to dashboard Cite

Throughout the history of art, the pose-as the holistic abstraction of the human body's expression-has proven to be a constant in numerous studies. However, due to the enormous amount of data that so far had to be processed by hand, its crucial role to the formulaic recapitulation of art-historical motifs since antiquity could only be highlighted selectively. This is true even for the now automated estimation of human poses, as domain-specific, sufficiently large data sets required for training computational models are either not publicly available or not indexed at a fine enough granularity. With the Poses of People in Art data set, we introduce the first openly licensed data set for estimating human poses in art and validating human pose estimators. It consists of 2,454 images from 22 art-historical depiction styles, including those that have increasingly turned away from lifelike representations of the body since the 19 th century. A total of 10,749 human figures are precisely enclosed by rectangular bounding boxes, with a maximum of four per image labeled by up to 17 keypoints; among these are mainly joints such as elbows and knees. For machine learning purposes, the data set is divided into three subsets-training, validation, and testing-, that follow the established JSON-based Microsoft Common Objects in Context (COCO) format, respectively. Each image annotation, in addition to mandatory fields, provides metadata from the art-historical online encyclopedia WikiArt. With this paper, we elaborate on the acquisition and constitution of the data set, address various application scenarios, and discuss prospects for a digitally supported art history. We show that the data set enables the comprehensive investigation of body phenomena in art, whether at the level of individual figures, which can thus be captured in their subtleties, or entire figure constellations, whose position, distance, or proximity to one another is considered.

show abstract

“…Recent evaluation benchmarks. Apart from the automatic metrics discussed above, multiple works involve human evaluation and propose their new evaluation benchmarks [14], [16], [63], [73], [80], [86], [87]. We summarize representative benchmarks in Table 2.…”

Section: Technical Evaluation Of Text-to-image Methodsmentioning

confidence: 99%

Generative Adversarial Networks for Diverse and Explainable Text-to-Image Generation

Zhang¹

View full text Add to dashboard Cite

The ArtBench Dataset: Benchmarking Generative Models with Artworks

Cited by 3 publications

References 27 publications

Genetic characteristics of Jiaji Duck by whole genome re-sequencing

Genetic characteristics of Jiaji Duck by whole genome re-sequencing

Poses of People in Art: A Data Set for Human Pose Estimation in Digital Art History

Generative Adversarial Networks for Diverse and Explainable Text-to-Image Generation

Contact Info

Product

Resources

About