Integrated Multiscale Domain Adaptive YOLO

TL;DR

This paper introduces MS-DAYOLO, a multiscale domain adaptive object detection framework based on YOLOv4, which improves detection accuracy across domains while maintaining real-time speed suitable for autonomous driving.

Contribution

The paper presents three novel deep learning architectures for domain adaptation integrated into YOLOv4, enhancing cross-domain object detection performance.

Findings

Significant improvement in detection accuracy on target domain datasets.

Order of magnitude faster than Faster R-CNN with comparable accuracy.

Effective multiscale domain adaptation for autonomous driving applications.

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. Building on our baseline multiscale DAYOLO framework, we introduce three novel deep learning architectures for a Domain Adaptation Network (DAN) that generates domain-invariant features. In particular, we propose a Progressive Feature Reduction (PFR), a Unified Classifier (UC), and an Integrated architecture. We train and test our proposed DAN architectures in conjunction with YOLOv4 using popular datasets. Our experiments show significant improvements in object detection performance when training YOLOv4 using the proposed MS-DAYOLO architectures and when tested on target data for autonomous driving applications. Moreover, MS-DAYOLO framework achieves an order of magnitude real-time speed improvement relative to Faster R-CNN solutions while providing comparable object detection performance.