cofga: A Dataset for Fine Grained Classification of Objects from Aerial Imagery

TL;DR

COFGA is a high-resolution aerial imagery dataset with detailed annotations, designed to advance fine-grained object classification research in overhead images, addressing challenges like small inter-class differences and intra-class variations.

Contribution

The paper introduces COFGA, a new open dataset with high-resolution aerial images and detailed labels, facilitating research in fine-grained classification of overhead objects.

Findings

COFGA contains 2,104 images with 14,256 annotated objects.

The dataset includes 37 labels across classes, subclasses, features, and colors.

Comparison shows COFGA's higher resolution and detailed annotations benefit fine-grained classification.

Abstract

Detection and classification of objects in overhead images are two important and challenging problems in computer vision. Among various research areas in this domain, the task of fine-grained classification of objects in overhead images has become ubiquitous in diverse real-world applications, due to recent advances in high-resolution satellite and airborne imaging systems. The small inter-class variations and the large intra class variations caused by the fine grained nature make it a challenging task, especially in low-resource cases. In this paper, we introduce COFGA a new open dataset for the advancement of fine-grained classification research. The 2,104 images in the dataset are collected from an airborne imaging system at 5 15 cm ground sampling distance, providing higher spatial resolution than most public overhead imagery datasets. The 14,256 annotated objects in the dataset were classified into 2 classes, 15 subclasses, 14 unique features, and 8 perceived colors a total of 37 distinct labels making it suitable to the task of fine-grained classification more than any other publicly available overhead imagery dataset. We compare COFGA to other overhead imagery datasets and then describe some distinguished fine-grain classification approaches that were explored during an open data-science competition we have conducted for this task.