Synthesizing Photorealistic Images with Deep Generative Learning

TL;DR

This thesis introduces new deep learning methods for photorealistic image synthesis, addressing tasks like image translation, completion, and scene decomposition, demonstrating superior results across multiple visual generation challenges.

Contribution

It presents five novel learning-based approaches for realistic image synthesis and extends some methods to related tasks like depth estimation.

Findings

Proposed approaches outperform existing methods in image synthesis quality.

Methods achieve high plausibility and visual realism in generated images.

Some approaches improve depth estimation accuracy.

Abstract

The goal of this thesis is to present my research contributions towards solving various visual synthesis and generation tasks, comprising image translation, image completion, and completed scene decomposition. This thesis consists of five pieces of work, each of which presents a new learning-based approach for synthesizing images with plausible content as well as visually realistic appearance. Each work demonstrates the superiority of the proposed approach on image synthesis, with some further contributing to other tasks, such as depth estimation.