Carousel: A High-Resolution Dataset for Multi-Target Automatic Image Cropping

Rafe Loya, Andrew Hamara, Benjamin Estell, Benjamin Kilpatrick, Andrew C. Freeman
11/6/2025
cs.CV

Official code: RafeLoya/carousel (3 stars, last commit 2025-07-18).

Abstract

Automatic image cropping is a method for maximizing the human-perceived quality of cropped regions in photographs. Although several works have proposed techniques for producing singular crops, little work has addressed the problem of producing multiple, distinct crops with aesthetic appeal. In this paper, we motivate the problem with a discussion on modern social media applications, introduce a dataset of 277 relevant images and human labels, and evaluate the efficacy of several single-crop models with an image partitioning algorithm as a pre-processing step. The dataset is available at https://github.com/RafeLoya/carousel.

View on arXivView PDF

Code Implementations(1)

RafeLoya/carouselOfficial100%

A High-Resolution Dataset for Multi-Target Automatic Image Cropping

31Jul 17, 20251 years ago

Similar Papers

Cropper: Vision-Language Model for Image Cropping through In-Context Learning

85% match2024

Learning Subject-Aware Cropping by Outpainting Professional Photos

85% match1 repo312023

Human-centric Image Cropping with Partition-aware and Content-preserving Features

85% match1 repo412022

Spatial-Semantic Collaborative Cropping for User Generated Content

84% match1 repo792024

Image Cropping under Design Constraints

83% match1 repo112023

Image Cropping with Spatial-aware Feature and Rank Consistency

82% match2023

Cite this paper

@article{loya2025carousel,
  title  = {Carousel: A High-Resolution Dataset for Multi-Target Automatic Image Cropping},
  author = {Rafe Loya and Andrew Hamara and Benjamin Estell and Benjamin Kilpatrick and Andrew C. Freeman},
  year   = {2025},
  eprint = {2511.04680},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2511.04680v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2511.04680.svg)](https://codeofpaper.com/paper/2511.04680)

Discussion