Abstract:Background: Image memorability refers to the phenomenon where certain images are more likely to be remembered than others. It is a quantifiable and intrinsic image attribute, defined as the likelihood of being remembered upon a single exposure. Despite advances in understanding human visual perception and memory, it is unclear what features contribute to an image's memorability. To address this question, we propose a deep learning-based computational modeling approach. Methods: We modeled the subjective experience of visual memorability using an autoencoder based on VGG16 Convolutional Neural Networks (CNNs). The model was trained on images for one epoch, to simulate the single-exposure condition used in human memory tests. We investigated the relationship between memorability and reconstruction error, assessed latent space representations distinctiveness, and developed a Gated Recurrent Unit (GRU) model to predict memorability likelihood. Interpretability analysis was conducted to identify key image characteristics contributing to memorability. Results: Our results demonstrate a significant correlation between the images memorability score and autoencoder's reconstruction error, and the robust predictive performance of its latent representations. Distinctiveness in these representations correlated significantly with memorability. Additionally, certain visual characteristics, such as strong contrasts, distinctive objects, and prominent foreground elements were among the features contributing to image memorability in our model. Conclusions: Images with unique features that challenge the autoencoder's capacity are inherently more memorable. Moreover, these memorable images are distinct from others the model has encountered, and the latent space of the encoder contains features predictive of memorability.

Modeling Visual Memorability Assessment with Autoencoders Reveals Characteristics of Memorable Images

Generating Memorable Images Based on Human Visual Memory Schemas

Feature representations useful for predicting image memorability

Predicting memorability of face photographs with deep neural networks

What Images are More Memorable to Machines?

How Much Training Data is Memorized in Overparameterized Autoencoders? An Inverse Problem Perspective on Memorization Evaluation

Defining Image Memorability using the Visual Memory Schema

Learning Computational Models of Video Memorability from Fmri Brain Imaging

Neural Encoding for Image Recall: Human-Like Memory

Changing the Image Memorability: From Basic Photo Editing to GANs

AMNet: Memorability Estimation with Attention

Image memorability enhances social media virality

Embedding Space Selection for Detecting Memorization and Fingerprinting in Generative Models

Controlling Memorability of Face Images

Media Memorability Prediction Based on Machine Learning

Predicting Image Memorability Through Adaptive Transfer Learning from External Sources

Seeing Eye to AI: Comparing Human Gaze and Model Attention in Video Memorability

What Makes Natural Scene Memorable?

Visual Image Classification Based on Auto-Associative Memory

Predicting Visual Memory Schemas with Variational Autoencoders

Memory for artwork is predictable