arXiv subjects

Artem Kozlov

Publications and source records attributed to Artem Kozlov.

2 recordsLinked to original sources

arXiv2026

Many-to-many. Usability challenges of entity reconciliation in art history and photographic studies

This article investigates challenges in reconciling heterogeneous records across cultural institutions, focusing on art historical photo archives within the PHAROS consortium. Through case studies, the study analyses reconciliation workflows and cataloguing traditions, with attention to institutional contexts, data granularities, and modelling strategies. Reconciliation is seldom a one-to-one operation. Ambiguities, incomplete data, shifting attributions, and varying practices shape outcomes. Strategies observed include the creation of anonymous or collective entities, the use of umbrella terms, the addition of uncertainty qualifiers, and reticence when ambiguity cannot be resolved. The article highlights the need to model uncertainty explicitly, offering a framework that connects technical reconciliation methods with institutional practices. Insights from PHAROS provide guidance for designing more robust, interoperable, and sustainable cultural heritage infrastructures.

cs.DL↗

arXiv2020

Working with scale: 2nd place solution to Product Detection in Densely Packed Scenes [Technical Report]

This report describes a 2nd place solution of the detection challenge which is held within CVPR 2020 Retail-Vision workshop. Instead of going further considering previous results this work mainly aims to verify previously observed takeaways by re-experimenting. The reliability and reproducibility of the results are reached by incorporating a popular object detection toolbox - MMDetection. In this report, I firstly represent the results received for Faster-RCNN and RetinaNet models, which were taken for comparison in the original work. Then I describe the experiment results with more advanced models. The final section reviews two simple tricks for Faster-RCNN model that were used for my final submission: changing default anchor scale parameter and train-time image tiling. The source code is available at https://github.com/tyomj/product_detection.

cs.CV↗