Efficient Multi-Crop Saliency Partitioning for Automatic Image Cropping

TL;DR

This paper presents an efficient method for automatic image cropping that extracts multiple salient regions simultaneously, improving over traditional single-region approaches by dynamically adjusting attention thresholds and avoiding full recomputation.

Contribution

It introduces a linear-time algorithm for multi-crop saliency partitioning that extends existing fixed aspect ratio cropping methods for better multi-region extraction.

Findings

Efficient extraction of multiple salient crops in linear time.

Dynamic adjustment of attention thresholds improves crop selection.

Potential for new datasets and benchmarks in saliency-aware cropping.

Abstract

Automatic image cropping aims to extract the most visually salient regions while preserving essential composition elements. Traditional saliency-aware cropping methods optimize a single bounding box, making them ineffective for applications requiring multiple disjoint crops. In this work, we extend the Fixed Aspect Ratio Cropping algorithm to efficiently extract multiple non-overlapping crops in linear time. Our approach dynamically adjusts attention thresholds and removes selected crops from consideration without recomputing the entire saliency map. We discuss qualitative results and introduce the potential for future datasets and benchmarks.