Robust Visual Perception in Open-World

Dear Colleagues,

With the rapid development of deep learning techniques, visual perception frameworks now often perform at a superhuman level in complex tasks (e.g., 3D object detection and tracking, panoptic segmentation, monocular depth estimation), which accelerates the deployment of computer vision systems in safety-critical applications, such as autonomous driving cars, robots, and intelligent surveillance. However, these systems often demonstrate performance drift when exposed to open-world settings, in which never-seen-before samples may show, or the data distribution may shift due to adverse weather. To resolve this, some methods, including data augmentation, domain adaptation, zero-shot learning, continual learning, etc., have been proposed, aimed at formulating a class-agnostic and domain-adaptive visual perception framework. However, some core challenges still remain unsolved, including how to accomplish robust perception when encountering novel classes in adverse condition, how to formulate a domain generalization model across different perception tasks and modalities, how to guarantee compatibility between robust perception and accurate decision making for autonomous systems, how to benchmark the existing robust visual perception algorithms in a more realistic and generalizable setting, how to quantify the uncertainty of the existing visual perception algorithms specifically designed for cross-class and cross-domain perception, and finally, how to leverage the multimodal pre-training model to formulate a better and more robust visual perception model.

This Special Issue will focus on collecting high-quality work presenting solutions and providing benchmarks for enabling robust visual perception in an open-world. This way, we seek to achieve a consensus on a rigorous framework to formulate and solve open-world visual perception problems and to characterize the properties that ensure the security of perceptual models deployed in autonomous systems.

Topics of interest include (but are not limited to):

• Visual perception with unknown classes
• Visual perception in adverse weather (e.g., hazy, rainy, nighttime, noisy, etc)
• Open-world visual perception in autonomous driving cars
• Open-world visual perception in embodied agents
• X-supervised learning in open-world
• Domain adaptation
• Lightweight yet robust visual perception models
• Visual perception with long-tailed data
• Pre-training visual models in open-world
• Knowledge-enriched visual perception in open-world

Dr. Guoqing Wang
Prof. Dr. Wen Li
Dr. Yalan Ye
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Electronics is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• visual perception
• robustness
• open-world
• autonomous systems