A research paper, now withdrawn, introduced CLIC (Contrastive Learning for Image Complexity), a framework utilizing contrastive learning to represent image complexity. The study proposed Random Crop and Mix (RCM) to generate diverse training samples from multi-scale local image crops, aiming to avoid manual annotation costs and human subjective biases. Experiments showed CLIC's performance was comparable to state-of-the-art supervised methods and could improve computer vision task performance. AI
IMPACT This research, though withdrawn, explored novel methods for image complexity representation that could inform future computer vision model development.
RANK_REASON The item is a withdrawn academic paper detailing a novel method for image complexity representation. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
