3D Clothed Human Reconstruction in the Wild (2207.10053v1)

Abstract: Although much progress has been made in 3D clothed human reconstruction, most of the existing methods fail to produce robust results from in-the-wild images, which contain diverse human poses and appearances. This is mainly due to the large domain gap between training datasets and in-the-wild datasets. The training datasets are usually synthetic ones, which contain rendered images from GT 3D scans. However, such datasets contain simple human poses and less natural image appearances compared to those of real in-the-wild datasets, which makes generalization of it to in-the-wild images extremely challenging. To resolve this issue, in this work, we propose ClothWild, a 3D clothed human reconstruction framework that firstly addresses the robustness on in-thewild images. First, for the robustness to the domain gap, we propose a weakly supervised pipeline that is trainable with 2D supervision targets of in-the-wild datasets. Second, we design a DensePose-based loss function to reduce ambiguities of the weak supervision. Extensive empirical tests on several public in-the-wild datasets demonstrate that our proposed ClothWild produces much more accurate and robust results than the state-of-the-art methods. The codes are available in here: https://github.com/hygenie1228/ClothWild_RELEASE.

• The paper presents a novel framework for reconstructing 3D clothed humans in wild, unconstrained settings.
• It leverages deep learning techniques to accurately capture intricate clothing details and body shapes from images.
• The approach demonstrates promising results with potential applications in virtual reality, gaming, and digital avatar creation.

Overview of ECCV Submission Guidelines

The document presented serves as a comprehensive guide for authors intending to submit papers to the European Conference on Computer Vision (ECCV). This guide delineates the specific formatting and submission requirements established for ECCV submissions, ensuring uniformity and compliance with the conference standards. This text aims to facilitate potential authors in adhering to these guidelines to the letter, thereby streamlining the submission process and enhancing the review experience.

Key Submission Requirements

The document imposes strict requirements on manuscript composition, focusing on aspects like language, length, and formatting. Papers must be written in English and restricted to 14 pages, excluding references. This length restriction is strictly enforced, with non-compliant submissions facing outright rejection. The authors are instructed to adhere to the default formatting styles and avoid modifications to font types or sizes, upholding the standardization of the submissions.

Anonymity is paramount in this process, with a strong emphasis on preserving it through double-blind reviewing. The authors are advised against revealing their identities through citations or supplementary materials that could be associated with them. The paper ID, vital for tracking and management throughout the review process, must be clearly stated on every page of the manuscript.

Dual and Double Submission Policies

A significant section of the document focuses on ECCV’s policies regarding dual and double submissions. It mandates that submissions should contain novel content not previously published or concurrently submitted elsewhere, emphasizing the conference's commitment to presenting pioneering research. Papers will undergo scrutiny for potential overlaps with existing work, ensuring the novelty of submissions.

Technical and Presentation Specifics

In terms of technical presentation, the guide provides elaborate instructions on the layout, typeface, and font sizes. Figures and tables should be included as floating objects, integrated seamlessly into the manuscript. These elements must meet the specified resolution standards and formatting style to maintain clarity and consistency across all submissions.

Additionally, the document addresses mathematical presentations, insisting on the numbering of equations and their correct citation in the manuscript to aid clarity and reference. The consistent and precise use of formatting extends to the creation of program code, figures, and other graphical elements within the paper.

Implications and Future Directions

The stringent guidelines outlined in this document reflect ECCV’s dedication to maintaining high academic standards, ensuring that the conference remains a prominent platform for cutting-edge research in computer vision. The emphasis on novel content, combined with meticulous formatting instructions, aims to improve the quality of submissions and, by extension, the papers presented at the conference.

As machine learning and artificial intelligence fields continue to evolve, similar guidelines will increasingly support the integrity and consistency of research dissemination at major conferences. These frameworks will likely adapt to technological advancements, further refining the review processes and elevating the standards of academic rigour.

In conclusion, the paper provides crucial insights for authors navigating the submission process for ECCV, laying down foundational principles for manuscript preparation. By following these comprehensive guidelines, researchers can enhance their prospects of acceptance and contribute meaningfully to the field of computer vision.