GAN-Control: Explicitly Controllable GANs (2101.02477v2)

Abstract: We present a framework for training GANs with explicit control over generated images. We are able to control the generated image by settings exact attributes such as age, pose, expression, etc. Most approaches for editing GAN-generated images achieve partial control by leveraging the latent space disentanglement properties, obtained implicitly after standard GAN training. Such methods are able to change the relative intensity of certain attributes, but not explicitly set their values. Recently proposed methods, designed for explicit control over human faces, harness morphable 3D face models to allow fine-grained control capabilities in GANs. Unlike these methods, our control is not constrained to morphable 3D face model parameters and is extendable beyond the domain of human faces. Using contrastive learning, we obtain GANs with an explicitly disentangled latent space. This disentanglement is utilized to train control-encoders mapping human-interpretable inputs to suitable latent vectors, thus allowing explicit control. In the domain of human faces we demonstrate control over identity, age, pose, expression, hair color and illumination. We also demonstrate control capabilities of our framework in the domains of painted portraits and dog image generation. We demonstrate that our approach achieves state-of-the-art performance both qualitatively and quantitatively.

LaTeX Author Guidelines for ICCV Proceedings

The document titled "LaTeX Author Guidelines for ICCV Proceedings" serves as a comprehensive guide for authors submitting manuscripts to the IEEE Computer Society Press for inclusion in the proceedings of the International Conference on Computer Vision (ICCV). The guidelines are crucial for maintaining consistency and ensuring the quality and readability of submitted papers.

Content Overview

The guidelines cover a broad range of technical specifications that authors must adhere to during the preparation of their manuscripts. These include:

• Abstract Formatting: Authors are instructed to format the abstract in fully-justified italicized text, a crucial detail for ensuring visual consistency across manuscripts. The abstract must immediately precede the main text with proper spacing and formatting.
• Submission Policies: The document details important submission policies such as dual submission guidelines, length restrictions for manuscripts, and the blind review process, emphasizing the importance of adhering to these rules to ensure fair and effective peer review.
• Technical Formatting: Detailed instructions on the formatting of figures, tables, page numbers, and sections ensure that manuscripts meet the technical and aesthetic standards set by the ICCV. Authors are instructed to maintain specific margins and use prescribed fonts for various sections of their paper.
• Mathematics and Citations: The guidelines emphasize the need for numbering all sections and equations within the manuscript to facilitate precise referencing and discussion. Authors are reminded of citation etiquette in the context of a blind review to maintain anonymity while still properly acknowledging previous work.
• Final Submission Requirements: Instructions on the final submission process, including the need for a signed IEEE copyright release form, underline the procedural requirements that must be fulfilled prior to publication.

Implications and Future Directions

The document primarily serves as a functional guide to ensure adherence to formatting standards, thereby contributing to the professional presentation of research findings. The structure and clarity that the guidelines impose are instrumental in presenting complex computer vision research in an accessible manner. Moreover, by enforcing strict formatting and citation rules, the guidelines help in preserving academic integrity and promoting a standardized review process.

Looking ahead, the guidelines could be instrumental in evolving the submission process as the field of computer vision continues to grow. A focus on digital submissions, automated formatting checks, and inclusive language guidelines could be integrated to further streamline the author experience and enhance the accessibility of academic publishing.

For researchers, adhering to such guidelines is critical not only for successful publication but also for contributing to a scholarly culture that values clarity, transparency, and academic rigor. As computer vision research advances, the structure provided by these guidelines will ensure that innovative findings are communicated effectively to a broader academic audience.