Developing Machine Learning Algorithms for Visual Recognition
In the realm of academic research, a dataset of abstract art images from the University of Bern has piqued the interest of many. However, a straightforward search for this dataset does not yield any publicly indexed results that directly mention or provide access to such a collection.
Given standard academic and data access practices, if this dataset exists, one would typically start by identifying the official project or research group webpage at the University of Bern that hosts or references this dataset. If the dataset is not openly hosted, reaching out to the corresponding author or principal investigator directly may be necessary to request access. This is common for datasets with institutional or privacy restrictions.
Registration or a formal data access agreement may also be required, depending on usage policies, similar to the process described for other research portals. Additionally, searching university repositories, institutional data archives, or platforms like Zenodo or figshare for any related dataset publications coming from University of Bern researchers could yield results.
Without explicit search results, the most reliable route would be to identify the dataset’s custodian at the University of Bern and request access by contacting them directly.
The dataset, if found, is said to contain 11,503 images of abstract art, though the time period of the abstract art is not specified. The collection features works from 53 different artists, and was released by researchers at the University of Bern. One image from the dataset is credited to DELAUNAY.
The dataset is intended for use in Switzerland and can be utilised to train image classification models. These models can be trained to recognise images with unusual structures, though the specific types of unusual structures the models are expected to recognise are not indicated. The dataset is specifically for abstract art images.
While the exact details of the dataset and its associated research group remain elusive, continued efforts to identify and access this valuable resource will undoubtedly contribute to the advancement of research in the field of abstract art and machine learning.
To progress in the research of abstract art images using AI, one might first explore the University of Bern's official project or research group webpage for potential leads on the dataset, as it is not publicly indexed. If the dataset's location remains unidentified, reaching out to the corresponding author or principal investigator could provide access to this resource, as they are likely its custodians. Once obtained, the dataset, comprising 11,503 images from 53 artists, could be utilized for training image classification models, with a focus on recognizing unusual structures specifically in abstract art images.
