# Presence-Only Geographical Priors for Fine-Grained Image Classification

## TL;DR

This paper introduces a spatio-temporal prior that leverages location and time data to improve fine-grained image classification, significantly enhancing accuracy by incorporating contextual cues often overlooked by traditional methods.

## Contribution

It presents a novel, efficient prior model trained on presence-only data that jointly captures object categories, their distributions, and biases, integrating this with image classifiers for better performance.

## Key findings

- Significant accuracy improvements when combining prior with classifiers
- Effective modeling of object distributions and biases from presence-only data
- Applicable across multiple challenging datasets

## Abstract

Appearance information alone is often not sufficient to accurately differentiate between fine-grained visual categories. Human experts make use of additional cues such as where, and when, a given image was taken in order to inform their final decision. This contextual information is readily available in many online image collections but has been underutilized by existing image classifiers that focus solely on making predictions based on the image contents. We propose an efficient spatio-temporal prior, that when conditioned on a geographical location and time, estimates the probability that a given object category occurs at that location. Our prior is trained from presence-only observation data and jointly models object categories, their spatio-temporal distributions, and photographer biases. Experiments performed on multiple challenging image classification datasets show that combining our prior with the predictions from image classifiers results in a large improvement in final classification performance.