How well a caption fits an image can be difficult to assess due to the subjective nature of caption quality. What is a caption? We investigate this problem by focusing on image-caption ratings and by generating high quality datasets from human feedback with gamification. We validate the datasets by showing a higher level of inter-rater agreement, and by using them to train custom machine learning models to predict new ratings. Our approach outperforms previous metrics - the resulting datasets are more easily learned and are of higher quality than other currently available datasets for image-caption rating.
Download full-text PDF |
Source |
|---|---|
| http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10962002 | PMC |
| http://dx.doi.org/10.1145/3544549.3585632 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Discrimination of Radiologists' Experience Level Using Eye-Tracking Technology and Machine Learning: Case Study.
JMIR Form Res
January 2025
Department of Psychology, The University of Texas at San Antonio, San Antonio, TX, United States.
Background: Perception-related errors comprise most diagnostic mistakes in radiology. To mitigate this problem, radiologists use personalized and high-dimensional visual search strategies, otherwise known as search patterns. Qualitative descriptions of these search patterns, which involve the physician verbalizing or annotating the order he or she analyzes the image, can be unreliable due to discrepancies in what is reported versus the actual visual patterns.
View Article and Find Full Text PDFDAU-Net: a novel U-Net with dual attention for retinal vessel segmentation.
Biomed Phys Eng Express
January 2025
Faculty of Information Technology, Beijing University of Technology, Beijing, People's Republic of China.
In fundus images, precisely segmenting retinal blood vessels is important for diagnosing eye-related conditions, such as diabetic retinopathy and hypertensive retinopathy or other eye-related disorders. In this work, we propose an enhanced U-shaped network with dual-attention, named DAU-Net, divided into encoder and decoder parts. Wherein, we replace the traditional convolutional layers with ConvNeXt Block and SnakeConv Block to strengthen its recognition ability for different forms of blood vessels while lightweight the model.
View Article and Find Full Text PDFDigital channel-enabled distributed force decoding via small datasets for hand-centric interactions.
Sci Adv
January 2025
Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Hong Kong, 999077, China.
Tactile interfaces are essential for enhancing human-machine interactions, yet achieving large-scale, precise distributed force sensing remains challenging due to signal coupling and inefficient data processing. Inspired by the spiral structure of and the processing principles of neuronal systems, this study presents a digital channel-enabled distributed force decoding strategy, resulting in a phygital tactile sensing system named PhyTac. This innovative system effectively prevents marker overlap and accurately identifies multipoint stimuli up to 368 regions from coupled signals.
View Article and Find Full Text PDFStandardized pipelines support and facilitate integration of diverse datasets at the Rat Genome Database.
Database (Oxford)
January 2025
Rat Genome Database, Department of Physiology, Medical College of Wisconsin, 8701 Watertown Plank Rd, Milwaukee, WI 53226, United States.
The Rat Genome Database (RGD) is a multispecies knowledgebase which integrates genetic, multiomic, phenotypic, and disease data across 10 mammalian species. To support cross-species, multiomics studies and to enhance and expand on data manually extracted from the biomedical literature by the RGD team of expert curators, RGD imports and integrates data from multiple sources. These include major databases and a substantial number of domain-specific resources, as well as direct submissions by individual researchers.
View Article and Find Full Text PDFProper 5'-3' cotranslational mRNA decay in yeast requires import of Xrn1 to the nucleus.
PLoS One
January 2025
Facultad de Biológicas, Instituto de Biotecnología y Biomedicina (BIOTECMED), Universitat de València, Burjassot, Spain.
The budding yeast Xrn1 protein shuttles between the nucleus, where it stimulates transcription, and the cytoplasm, where it executes the major cytoplasmic mRNA decay. In the cytoplasm, apart from catalyzing 5'→3' decay onto non translated mRNAs, Xrn1 can follow the last translating ribosome to degrade the decapped mRNA template, a process known as "cotranslational mRNA decay". We have previously observed that the import of Xrn1 to the nucleus is required for efficient cytoplasmic mRNA decay.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!