Visual Attention: Deep Rare Features

TL;DR

DeepRare2019 is a non-training, fast, and generic model that effectively detects surprising or unusual data in images, outperforming other models across various eye-tracking datasets.

Contribution

It introduces a training-free, efficient deep feature-based model that consistently ranks among the top in detecting interesting image data across diverse datasets.

Findings

DeepRare2019 does not require training and runs in less than a second per image.

It outperforms other models on multiple eye-tracking datasets.

The model demonstrates high regularity and genericity across different metrics.

Abstract

Human visual system is modeled in engineering field providing feature-engineered methods which detect contrasted/surprising/unusual data into images. This data is "interesting" for humans and leads to numerous applications. Deep learning (DNNs) drastically improved the algorithms efficiency on the main benchmark datasets. However, DNN-based models are counter-intuitive: surprising or unusual data is by definition difficult to learn because of its low occurrence probability. In reality, DNNs models mainly learn top-down features such as faces, text, people, or animals which usually attract human attention, but they have low efficiency in extracting surprising or unusual data in the images. In this paper, we propose a model called DeepRare2019 (DR) which uses the power of DNNs feature extraction and the genericity of feature-engineered algorithms. DR 1) does not need any training, 2) it takes less than a second per image on CPU only and 3) our tests on three very different eye-tracking datasets show that DR is generic and is always in the top-3 models on all datasets and metrics while no other model exhibits such a regularity and genericity. DeepRare2019 code can be found at https://github.com/numediart/VisualAttention-RareFamily