CLIC: Contrastive Learning Framework for Unsupervised Image Complexity Representation

TL;DR

CLIC is an unsupervised contrastive learning framework that effectively captures image complexity features without manual labels, improving downstream task performance and overcoming limitations of traditional metrics.

Contribution

The paper introduces CLIC, a novel unsupervised contrastive learning method with a unique sample selection strategy and complexity-aware loss, for quantifying image complexity.

Findings

CLIC achieves competitive results with limited labeled data.

Pretraining and application are free from subjective bias.

Enhances performance in downstream computer vision tasks.

Abstract

As a fundamental visual attribute, image complexity significantly influences both human perception and the performance of computer vision models. However, accurately assessing and quantifying image complexity remains a challenging task. (1) Traditional metrics such as information entropy and compression ratio often yield coarse and unreliable estimates. (2) Data-driven methods require expensive manual annotations and are inevitably affected by human subjective biases. To address these issues, we propose CLIC, an unsupervised framework based on Contrastive Learning for learning Image Complexity representations. CLIC learns complexity-aware features from unlabeled data, thereby eliminating the need for costly labeling. Specifically, we design a novel positive and negative sample selection strategy to enhance the discrimination of complexity features. Additionally, we introduce a complexity-aware loss function guided by image priors to further constrain the learning process. Extensive experiments validate the effectiveness of CLIC in capturing image complexity. When fine-tuned with a small number of labeled samples from IC9600, CLIC achieves performance competitive with supervised methods. Moreover, applying CLIC to downstream tasks consistently improves performance. Notably, both the pretraining and application processes of CLIC are free from subjective bias.