From Fog to Failure: The Unintended Consequences of Dehazing on Object Detection in Clear Images

TL;DR

This paper investigates how dehazing techniques, beneficial in foggy conditions, can unexpectedly impair object detection in clear images, emphasizing the importance of selective preprocessing in hybrid detection pipelines.

Contribution

It introduces a multi-stage framework combining lightweight detection and spatial attention-based dehazing, and analyzes its unintended negative effects on clear images.

Findings

Dehazing improves foggy image detection but degrades clear image detection.

Selective preprocessing is crucial for balanced detection performance.

Hybrid pipelines require careful design to avoid unintended consequences.

Abstract

This study explores the challenges of integrating human visual cue-based dehazing into object detection, given the selective nature of human perception. While human vision adapts dynamically to environmental conditions, computational dehazing does not always enhance detection uniformly. We propose a multi-stage framework where a lightweight detector identifies regions of interest (RoIs), which are then improved via spatial attention-based dehazing before final detection by a heavier model. Though effective in foggy conditions, this approach unexpectedly degrades the performance on clear images. We analyze this phenomenon, investigate possible causes, and offer insights for designing hybrid pipelines that balance enhancement and detection. Our findings highlight the need for selective preprocessing and challenge assumptions about universal benefits from cascading transformations.