一、方法
-
Approach 二、考虑
-
Consider 三、本质上
-
In spirit 四、据我们所知、众所周知
-
To our best knowledge 五、解决(+问题)
-
Address 六、基本的(+思想)
-
Basic 七、关键的(+部件)
-
Key 八、问题(替换problem)
-
Issue 九、使用(+model、方法)
-
Employ 十、提出(+方法、model、算法)
-
Propose 十一、(视觉+)场景
-
Scenario 十二、(图表+)描述、刻画
-
Depict 十三、(矩阵、向量+)连接
-
Concatenate 十四、(特征和图像之间的+)匹配程度
-
degree-of-match 十五、分别的、各自的
-
Respectively 十六、专门的、定制的、特制的
-
Tailored
-
Tailor-made 十七、简单的、易懂的、不复杂的
-
Straightforward 十八、表达、表示、如…所示、说明、证明
-
Demonstrate
-
Denote
-
represent 一、方法
-
Approach the approach is tailored made for pizza image generation with limited kinds of ingredients and cooking methods being considered. 这种方法是专门为比萨饼图像生成而设计的,只考虑了有限种类的配料和烹饪方法。
二、考虑
-
Consider Different from this paper, the issue of how the cause of a cooking step results in a new effect for ingredients is not considered. 与本文不同的是,没有考虑烹饪步骤的原因如何对配料产生新影响的问题
三、本质上
-
In spirit Similar in spirit as other GANs [19, 20, 16], CookGAN progressively upsamples image from low to high resolution. 在本质上类似于其他GAN[19,20,16],CookGAN逐步提高图像从低分辨率到高分辨率。
四、据我们所知、众所周知
-
To our best knowledge To our best knowledge, the high-resolution food image generation considering causality is still an unexplored problem. 据我们所知,考虑因果关系的高分辨率食品图像生成仍然是一个尚未探索的问题。
五、解决(+问题)
-
Address This paper addresses the problem of text-to-image synthesis from a new perspective. 本文从一个新的角度研究了文本到图像的合成问题。
六、基本的(+思想)
-
Basic The basic idea of CookGAN is to progressively upsample the small-size image up to a resolution of 256×256. CookGAN的基本思想是逐步向上采样小尺寸图像,使其分辨率达到256×256。
七、关键的(+部件)
-
Key The key component of CookGAN is Cooking Simulator. CookGAN的关键部件是烹饪模拟器
八、问题(替换problem)
-
Issue CookGAN addresses four issues in generating causality realistic food image. 提出问题
九、使用(+model、方法)
-
Employ By employing CGAN, 通过使用CGAN
十、提出(+方法、model、算法)
-
Propose We propose a new network architecture 我们提出了一种新的网络结构。
十一、(视觉+)场景
-
Scenario The key component of CookGAN is Cooking Simulator (Section 3.2), which generates causeand-effect visual scenario. CookGAN的关键部件是烹饪模拟器(第3.2节),它生成因果视觉场景。
十二、(图表+)描述、刻画
-
Depict Figure 2 depicts the network of Cooking Simulator. 图2描述了烹饪模拟器的网络。
十三、(矩阵、向量+)连接
-
Concatenate Initially, an embedded feature ϕr is concatenated with random noise z sampled from a Gaussian distribution Z ∼N(0,I). 最初,嵌入特征ɕr与从高斯分布z∼N(0,I)采样的随机噪声z相连。
十四、(特征和图像之间的+)匹配程度
-
degree-of-match The latter term is to evaluate the degree-of-match between an generated image and its embedded feature initially extracted from R2GAN. 后一项是评估生成的图像与其最初从R2GAN中提取的嵌入特征之间的匹配程度。
十五、分别的、各自的
-
Respectively R2GAN embeds the features extracted from ingredients and cooking instructions respectively into a latent space compatible with food images. R2GAN将分别从配料和烹饪说明中提取的特征嵌入到与食品图像兼容的潜在空间中。
十六、专门的、定制的、特制的
-
Tailored the approach is tailored made for pizza image generation with limited kinds of ingredients and cooking methods being considered. 这种方法是专门为比萨饼图像生成而设计的,只考虑了有限种类的配料和烹饪方法。
-
Tailor-made The Cooking Simulator is a tailor-made sub-network to implicitly model cause-and-effect visual change in cooking process. 烹饪模拟器是一个特制的子网络,用于隐式地模拟烹饪过程中的视觉变化。
十七、简单的、易懂的、不复杂的
-
Straightforward Extension for dishes beyond pizza is not straightforward. 比萨以外的菜肴的扩展并不简单。
十八、表达、表示、如…所示、说明、证明
-
Demonstrate As demonstrated by R2GAN, the embedded feature, denoted as ϕr, can be used to generate a thumbnail image of resolution 64×64. 如R2GAN所示,所述嵌入特征可用于生成分辨率为64×64的缩略图。
-
Denote Denote ϕing = {ϕingm}M m=1 ∈ RM×ding as a list of ingredients, where ϕingm represents the ding-dimensional vector of mth ingredient in the list. 表示ψing={ψingm}M M=1∈RM×ding作为一个成分列表,其中ɕingm表示列表中第M个成分的d维向量。
-
represent (在上面denote例句中出现)
另外:
-
furthermore
-
moreover
-
in addition / in addition to that
-
additionally