Vocabulary and Phrase in Paper of Computer Vision (Updating)

18 篇文章 4 订阅
3 篇文章 0 订阅

Vocabulary and Phrase in Paper of Computer Vision

Sec 1 RCNN

v&pcharacteristictranslationusage
as measured on正如…所衡量的as measured on the canonical PASCAL VOC dataset
plateauvi达到平衡;达到稳定阶段B has plateaued in the last few years.某领域近年来趋于平稳
best-performingadj最佳效果的用来形容某领域中效果最佳的方法
scalableadj可扩展的用于形容一种可扩展的算法
more than 30% relative to …相比于…多了30%more than 30% relative to the previous best result
scarceadj;adv不足的;稀有的形容某物很少
domain-specific特定领域的
yieldvi;vt导致;屈服同lead
boostn;v提升;促进;增加同improvement
by a large margin以一个很高的状态A outperforms B by a large margin; A极大地超越了B
canonicaladj权威性的the canonical visual recognition task 权威性的视觉识别工作
it is generally acknowledged that…是被认可的共识形容某种趋势,方法的优点
minor variants of successful methods对成功方法的微小改进
… see heavy use …得到大量使用CNNs saw heavy use in the 1990s;CNN在90年代得到了大量使用
rekindlev重新点燃
vigorouslyadv强有力地;茁壮地;强烈地
frame … as …将…定义为…
may not fare well in practice可能实际表现并不理想
alternativeadj;n替代的;替代品
… dub …v…给…起绰号
head-to-head comparison of … and ……和…的直接较量
category-independentadj类别无关的
warpv扭曲,使弯曲warp all pixels in a tight bounding box around it to the required size;弯曲所有像素使其达到所需大小
time spentn时间消耗
tightly enclose密切地紧贴着形容某region将某物体最小包围
elucidatev阐明同depict
stems from源于The inefficiency stems from the fact that …
One concern over this strategy对这种策略的担忧
streamlinev,n,adj流线型,简单化,合理化,流线a streamlined training process
siblingn兄弟姐妹A Fast R-CNN network has two sibling output layers.
by convention按照惯例By convention the catch-all background class is labeled u = 0.
catch-all所有的
notionn概念;见解;打算there is no notion of a ground-truth bounding box 没有gt 包围盒的概念

Sec 2 Feature Selective Anchor-Free Module for Single-Shot Object Detection(CVPR2019)

v&pcharacteristictranslationusage
plug A into B将A嵌入到B中A is plugged into B
limitationn限制,缺点,弱项It addresses two limitations
brought up by由…带来的

Sec 3 Small Obstacle Avoidance Based on RGB-D Semantic Segmentation(ICCV2019 Workshop)

v&pcharacteristictranslationusage
local path planning局部路径规划
morphological processing形态学处理
monochromeadj单色的,黑白的monochrome image

Sec 4 MultiNet: Real-time Joint Semantic Reasoning for Autonomous Driving(IV 2018)

v&pcharacteristictranslationusage
prominentlyadv显著地;重要地the input features one object, usually centred prominently in the image
serve as用作The purpose of this encoder is to serve as high quality baseline
joint inference联合推理(多任务神经网络)This first classification encoder cannot be used for joint inference with segmentation and detection
featuren,v使突出,容貌,特点the input features one object, usually centred prominently in the image
marry A with B将A特性赋予B

Sec 5 A Comprehensive Survey on Graph Neural Networks

v&pcharacteristictranslationusage
revolutionizevt使彻底变革Deep learning has revolutionized many machine learning tasks in recent years.
taxonomyn分类学,分类系统

Other

v&pcharacteristictranslationusage
prevailingadj占优势的;主要的;流行的;普遍的prevailing methods
tacklev,n解决,应付tackle the task
symptomn征兆;症状a typical symptom is distorted and blurry reconstruction of object boundaries.
revisitv,n重游;再访;重临revisiting Single Image Depth Estimation表达对某物的重新探讨
substantiallyadv可观地considerablely
be superior to比…更优This method is superior to direct point cloud interpolation methods in term of feasibility and efficiency.
lightweightadj n轻量级的the lightweight model
affluentadj富裕的;丰富的;流畅的同plentiful
filter out过滤掉effectively filter out the incorrect bounding boxes
resortn,v使用,凭借,诉诸,手段resort to the following procedure
maneuvern,vi,vt调遣,机动,策略the proposed method can be used to guide the maneuver of the robot
inherentadj固有的,内在的we analyze a problem inherent to any homography-based approach
resultantadj因而发生的, 必然产生的Due to the reduced degree of freedom, the resultant algorithm …
stringentadj严格的,严厉的,紧缩的,短缺的there is no stringent requirement on camera calibration
In theory理论上说
stablyadv坚固地;安定地;平稳地
be on par with…与…相当The results are on par with supervised methods
laboriousadj.勤劳的;困难的, 费力的Compared with laborious pixel-wise dense labeling
scribblen,v杂文,乱写,滥写
mitigatev缓解,使缓和
sb note A B某人将A定义为BWe note L a ray from C
iff当且仅当If and only if
counterpartn用于指代作用相当的物体Contrastive learning benefits from larger batch sizes and longer training compared to its supervised counterpart。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值