Cross-domain object detection using unsupervised image translation. (15th April 2022)

• Record Type: Journal Article
• Authors: Arruda, Vinicius F.; Berriel, Rodrigo F.; Paixão, Thiago M.; Badue, Claudine; De Souza, Alberto F.; Sebe, Nicu; Oliveira-Santos, Thiago
• Is Part Of: Expert systems with applications. Volume 192(2022)
• Publication Date: 2022-04-15
• Subjects: Unsupervised Domain Adaptation -- Object detection -- Generative Adversarial Networks -- Unpaired image-to-image translation -- Style-transfer; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2021.116334
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: Unsupervised domain adaptation for object detection addresses the adaption of detectors trained in a source domain to work accurately in an unseen target domain. Recently, methods approaching the alignment of the intermediate features proven to be promising, achieving state-of-the-art results. However, these methods are laborious to implement and hard to interpret. Although promising, there is still room for improvements to close the performance gap toward the upper-bound (when training with the target data). In this work, we propose a method to generate an artificial dataset in the target domain to train an object detector. We employed two unsupervised image translators (CycleGAN and an AdaIN-based model) using only annotated data from the source domain and non-annotated data from the target domain. Our key contributions are the proposal of a less complex yet more effective method that also has an improved interpretability. Results on real-world scenarios for autonomous driving show significant improvements, outperforming state-of-the-art methods in most cases, further closing the gap toward the upper-bound. Highlights: A simple yet effective method for detecting objects on unsupervised domain adaptation. Artificially generated images are useful for unsupervised domain adaptation. An extensive comparison with the state-of-the-art is provided. Experiments in three scenarios: synthetic data, adverse weather, and cross-camera.