Examples

Misc

w.l.o.g. = without loss of generality.
... be prohibitive of ...
... be prohibitively computational expensive ...

Abstract

It is of increasing importance for ... and draws increasing attention.
Further ablation study and analysis gives more insights into the proposed method and demonstrate the generalization capability and stability of our model.
Experiments on the ... dataset show that xxx outperforms the state-of-the-arts in xxx, especially for ...

Intro

... remains an important open challenge in ...
Recently, there has been progress in ...
Due to the difficulty of the task, these methods still have significant limitations. In particular, early efforts focused on ...
We are motivated by these observations to ...
Along the line of research that , ...
Could ... ? Prior work is encouraging. [OpenAI CLIP]
... is one of the central tasks in 3D computer vision with many applications. For example, ...
... is a fundamental problem in computer vision and computer graphics.
... has recently become a highly promising alternative to classical xxx approaches due to its ...
The ability to ... is a long-standing goal of computer vision and graphics.
... is a core element in diverse fields of research, including ....
We study alternative ways for ... to improve ... beyond ...
The benefits from the ... scheme lie in two-fold. On the one hand, ... On the other hand, ....
by ..., we make it possible to ..., and thus have the best of both worlds, i.e., ... and ....
whereas previous works reach high performance on standard ... benchmarks, in challenging real-world conditions, it is desirable to involve ...
for ..., there are two dominant paradigms. The first one focuses on ..., Another mainstream solution ...
while state-of-the-art A methods handle xxx well, B methods either ignore these cases altogether or only deliver a weak performance.
These methods fall short of generating novel samples of a subject while preserving fidelity.
We do not impose any restrictions on input image capture settings.
In addition, language drift is a common problem in language models, and manifests itself in text-to-image diffusion models as well.
In contrast, ...

Contributions

We present a new approach capable of ... from ...
We propose XXX, a novel ..., which ...
We demonstrate significant performance improvements of our approach compared to prior work.

We briefly describe related work in major lines of research: ...
We are motivated to extend this line of research to a ... setting.
This severely curtails their flexibility and limits their ... capabilities.
The most related line of research is ...
Before the age of deep learning, ...
... have been shown to allow for ...
.... has gained popularity in ... To overcome ..., several works ...
Several works investigate how ... While many approaches use ..., we focus on ...
Compared to the xxx approaches, wherein the ... is often suboptimal, yyy approaches are more straightforward and have achieved promising performance.

Methods

(directly describe methods) In this section, we briefly summarize A, and then describe how we extend it to B.
(first describe problems) Given , our goal is to ...
We emphasize that what is common across this line of work is not but . [CLIP]
Our goal is .... In the following, we discuss the main components of our method. First, ... (See Sec~\ref{}). Next, ... Finally, ....

Experiments

The improvements potentially come from the design of ... in our method, which can act as a ...

Appendix

The proof is given in the Appendix.
The detailed derivation of this formula is given in the Appendix.

junction

In this section we investigate the relationship between ...
In this section, we first verify the efficacy of ..., and then show that ...
In particular, we show that ..., thus motivating ...
For the sake of simplicity and w.l.o.g, hereafter we ...
Let us consider a ...
Equivalently, ...

Examples

Misc

Abstract

Intro

Contributions

Related work

Methods

Experiments

Appendix

junction