Video Object Segmentation: From Semi-supervised to Unsupervised

Dear Colleagues,

Video object segmentation usually concentrates on semi-supervised, weak supervised, and unsupervised modes. Semi-supervised video object segmentation: the current video object segmentation model usually learns in a fully supervised manner, requiring equivalence between input training samples and their expected outputs after changes. The standard method used to evaluate learning outcomes follows the empirical risk loss minimization formula, which evaluates the differences between estimated values and prior knowledge related to video segmentation, but at the cost of requiring a large amount of well-labeled data. Weak supervised video object segmentation: this model is employed using labels that are easier to annotate, such as circles, bounding boxes, or graffiti, and using limited samples to approximate the mapping equation. Unsupervised video object segmentation: when no manual annotations are provided in the initial frame, the video object segmentation used is called unsupervised video object segmentation. Unsupervised video object segmentation methods focus more on mining the intrinsic attributes of the video (such as cross-frame consistency) and propagating them throughout the entire video.

So far, the semi-supervised learning method based on deep learning has played a dominant role in the field of video object segmentation. However, exploring this task in a weakly supervised or unsupervised manner is more attractive not only because these can reduce the burden of manual annotation, but also, by exploring prior information, one can gain a deeper understanding of the essence of video object segmentation tasks. The scope of this topic includes but is not limited to:

1. Video object detection, identification, recognition, tracking, and segmentation.
2. Video analysis and tracking.
3. Image and video enhancement algorithms to improve the quality of video object tracking.
4. Computational photography and imaging for advanced object detection and tracking.
5. Depth estimation and three-dimensional reconstruction for augmented reality (AR) and/or advanced driver assistance systems (ADAS).
6. Learning data representation from video based on supervised, unsupervised, and semi-supervised learning.
7. Dataset and performance evaluation, person re-identification, vehicle re-identification.
8. Human behavior detection, human pose estimation, and tracking.
9. Video object surveillance and monitoring.

Prof. Dr. Kaihua Zhang
Guest Editor

Dr. Jiaqing Fan
Guest Editor Assistant

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. Mathematics 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 2600 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.

• object tracking
• image segmentation