Benchmarking performance of object detection under image distortions in an uncontrolled environment

TL;DR

This paper evaluates the robustness of object detection algorithms under various image distortions, introduces a new distortion generation strategy, and provides a dataset to improve and assess detection performance in uncontrolled environments.

Contribution

It presents a novel distortion generation method applied to MS-COCO, a new dataset for robustness evaluation, and demonstrates improved detection robustness through training on this dataset.

Findings

Training on the proposed distorted dataset improves robustness by 31.5%.

The new distortion strategy enhances object detection performance under severe distortions.

The publicly available dataset enables more reliable evaluation of detection algorithms.

Abstract

The robustness of object detection algorithms plays a prominent role in real-world applications, especially in uncontrolled environments due to distortions during image acquisition. It has been proven that the performance of object detection methods suffers from in-capture distortions. In this study, we present a performance evaluation framework for the state-of-the-art object detection methods using a dedicated dataset containing images with various distortions at different levels of severity. Furthermore, we propose an original strategy of image distortion generation applied to the MS-COCO dataset that combines some local and global distortions to reach much better performances. We have shown that training using the proposed dataset improves the robustness of object detection by 31.5\%. Finally, we provide a custom dataset including natural images distorted from MS-COCO to perform a more reliable evaluation of the robustness against common distortions. The database and the generation source codes of the different distortions are made publicly available