Multiscale Domain Adaptive YOLO for Cross-Domain Object Detection

TL;DR

This paper introduces MS-DAYOLO, a multiscale domain adaptive framework for YOLOv4 that enhances object detection across different domains, especially under challenging weather conditions for autonomous driving.

Contribution

The paper proposes a novel multiscale domain adaptation approach integrated into YOLOv4, improving cross-domain object detection performance.

Findings

Significant performance improvements on target datasets with challenging weather conditions.

Effective generation of domain-invariant features across multiple scales.

Enhanced robustness of YOLOv4 in cross-domain scenarios.

Abstract

The area of domain adaptation has been instrumental in addressing the domain shift problem encountered by many applications. This problem arises due to the difference between the distributions of source data used for training in comparison with target data used during realistic testing scenarios. In this paper, we introduce a novel MultiScale Domain Adaptive YOLO (MS-DAYOLO) framework that employs multiple domain adaptation paths and corresponding domain classifiers at different scales of the recently introduced YOLOv4 object detector to generate domain-invariant features. We train and test our proposed method using popular datasets. Our experiments show significant improvements in object detection performance when training YOLOv4 using the proposed MS-DAYOLO and when tested on target data representing challenging weather conditions for autonomous driving applications.