




版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請(qǐng)進(jìn)行舉報(bào)或認(rèn)領(lǐng)
文檔簡(jiǎn)介
1、16-721: Learning-based Methods in VisionStaff:Instructor: Alexei (Alyosha) Efros (efroscs), 4207 NSHTA: Jean-Francois Lalonde (jlalondecs), A521 NSHWeb Page:/efros/courses/LBMV07/TodayIntroductionWhy This Course?Administrative stuffOverview of the courseImage DatasetsProjects / ChallengesA bit about
2、 meAlexei (Alyosha) EfrosRelatively new faculty (RI/CSD)Ph.D 2003, from UC Berkeley (signed by Arnie!)Research Fellow, University of Oxford, 03-04TeachingI am still learningThe plan is to have fun and learn cool things, both you and me!Social warning: I dont see well ResearchVision, Graphics, Data-d
3、riven “stuff”PhD Thesis on Texture and Action SynthesisAntonio Criminisis son cannot walk but he can flySmart Erase button in Microsoft Digital Image Pro:Why this class?The Old Days:1. Graduate Computer Vision2. Advanced Machine PerceptionWhy this class?The New and Improved Days:1. Graduate Computer
4、 Vision2. Advanced Machine PerceptionPhysics-based Methods in VisionGeometry-based Methods in VisionLearning-based Methods in VisionDescribing Visual Scenes using Transformed Dirichlet Processes. E. Sudderth, A. Torralba, W. Freeman, and A. Willsky. NIPS, Dec. 2005. The Hip & Trendy LearningLearning
5、 as Last ResortLearning as Last Resortfrom Sinha and Adelson 1993EXAMPLE: Recovering 3D geometry from single 2D projection Infinite number of possible solutions! Learning-based Methods in VisionThis class is about trying to solve problems that do not have a solution! Dont tell your mathematician fri
6、neds!This will be done using Data:E.g. what happened before is likely to happen againGoogle Intelligence (GI): The AI for the post-modern world!Why is this even useful?Even a decade ago at ICCV99 Faugeras claimed it wasnt!The Vision Story Begins“What does it mean, to see? The plain mans answer (and
7、Aristotles, too). would be, to know what is where by looking.” - David Marr, Vision (1982) Vision: a split personality“What does it mean, to see? The plain mans answer (and Aristotles, too). would be, to know what is where by looking. In other words, vision is the process of discovering from images
8、what is present in the world, and where it is.” Answer #1: pixel of brightness 243 at position (124,54) and depth .7 metersAnswer #2: looks like bottom edge of whiteboard showing at the top of the imageWhich Do we want?Is the difference just a matter of scale? depth mapMeasurement vs. PerceptionBrig
9、htness: Measurement vs. PerceptionBrightness: Measurement vs. PerceptionProof!Lengths: Measurement vs. Perceptionhttp:/www.michaelbach.de/ot/sze_muelue/index.html Mller-Lyer IllusionVision as Measurement DeviceReal-time stereo on MarsStructure from MotionPhysics-based VisionVirtualized Realitybut wh
10、y do Learning for Vision?“What if I dont care about this wishy-washy human perception stuff? I just want to make my robot go!” Small Reason: For measurement, other sensors are often better (in DARPA Grand Challenge, vision was barely used!)For navigation, you still need to learn!Big Reason: The goal
11、s of computer vision (what + where) are in terms of what humans care about.So what do humans care about?slide by Fei Fei, Fergus & Torralba Verification: is that a bus?slide by Fei Fei, Fergus & Torralba Detection: are there cars?slide by Fei Fei, Fergus & Torralba Identification: is that a picture
12、of Mao?slide by Fei Fei, Fergus & Torralba Object categorizationskybuildingflagwallbannerbuscarsbusfacestreet lampslide by Fei Fei, Fergus & Torralba Scene and context categorization outdoor city traffic slide by Fei Fei, Fergus & Torralba Rough 3D layout, depth orderingChallenges 1: view point vari
13、ationMichelangelo 1475-1564Challenges 2: illuminationslide credit: S. UllmanChallenges 3: occlusionMagritte, 1957 Challenges 4: scaleslide by Fei Fei, Fergus & Torralba Challenges 5: deformationXu, Beihong 1943Challenges 6: background clutterKlimt, 1913Challenges 7: object intra-class variationslide
14、 by Fei-Fei, Fergus & Torralba Challenges 8: local ambiguityslide by Fei-Fei, Fergus & Torralba Challenges 9: the world behind the image In this course, we will:Take a few baby stepsGoalsRead some interesting papers togetherLearn something new: both you and me!Get up to speed on big chunk of vision
15、researchunderstand 70% of CVPR papers!Use learninig-based vision in your own workTry your hand in a large vision projectLearn how to speakLearn how think critically about papersCourse OrganizationRequirements:Paper Presentations (50%)Paper PresenterPaper EvaluatorClass Participation (20%)Keep annota
16、ted bibliography Ask questions / debate / flight / be involved!Final Project (30%)Do something with lots of data (at least 500 images)Groups of 1 or 2Paper AdvocatePick a paper from list That you like and willing to defendSometimes I will make you do two papers, or backgroundMeet with me before star
17、ting, to talk about how to present the paper(s)Prepare a good, conference-quality presentation (20-45 min, depending on difficulty of material) Meet with me again 2 days before class to go over the presentationOffice hours at end of each classPresent and defend the paper in front of classPaper Evalu
18、ator For some papers, we will have EvaluatorsSign up for a paper you find interestingGet the code online (or implement if easy)Run it on a toy problem, play with parametersRun it on a new datasetPrepare short 10-15 min presentation detailing resultsDiscuss the paper criticallyClass ParticipationKeep
19、 annotated bibliography of papers you read (always a good idea!). The format is up to you. At least, it needs to have:Summary of key pointsA few Interesting insights, “aha moments”, keen observations, etc.Weaknesses of approach. Unanswered questions. Areas of further investigation, improvement.Submit your thoughts for current paper
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請(qǐng)下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請(qǐng)聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁(yè)內(nèi)容里面會(huì)有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
- 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
- 5. 人人文庫(kù)網(wǎng)僅提供信息存儲(chǔ)空間,僅對(duì)用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對(duì)用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對(duì)任何下載內(nèi)容負(fù)責(zé)。
- 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請(qǐng)與我們聯(lián)系,我們立即糾正。
- 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶因使用這些下載資源對(duì)自己和他人造成任何形式的傷害或損失。
最新文檔
- 甘肅卷-2025屆高考?xì)v史4月模擬預(yù)測(cè)卷(解析版)
- 2023年甘肅省煙草專賣局(公司)應(yīng)屆高校畢業(yè)生招聘124人筆試參考題庫(kù)附帶答案詳解
- 2025至2030年中國(guó)玩具燈籠數(shù)據(jù)監(jiān)測(cè)研究報(bào)告
- 2025至2030年中國(guó)汽車子板內(nèi)襯數(shù)據(jù)監(jiān)測(cè)研究報(bào)告
- 2025至2030年中國(guó)普通毛紗數(shù)據(jù)監(jiān)測(cè)研究報(bào)告
- 2025年中國(guó)胸圍尺市場(chǎng)調(diào)查研究報(bào)告
- 河南省洛陽市強(qiáng)基聯(lián)盟2024-2025學(xué)年高一下學(xué)期3月月考?xì)v史試題 (原卷版+解析版)
- 飲品店加盟店培訓(xùn)
- 骨科疾病的綜合治療方案
- 消防知識(shí)講座馬克私
- 袁家村鄉(xiāng)村旅游運(yùn)營(yíng)模式簡(jiǎn)析講解講解
- 攝影實(shí)踐教程全套教學(xué)課件
- MOOC 數(shù)據(jù)結(jié)構(gòu)-西北大學(xué) 中國(guó)大學(xué)慕課答案
- 品質(zhì)部組織架構(gòu)圖構(gòu)
- 2023年湖北宜昌高新區(qū)社區(qū)專職工作人員(網(wǎng)格員)招聘考試真題及答案
- 小學(xué)數(shù)學(xué)課堂“教學(xué)評(píng)一致性”的有效實(shí)踐
- 認(rèn)識(shí)常用電子元件圖解
- 消防信息化建設(shè)-智慧消防方案
- 2024年全球無創(chuàng)血糖監(jiān)測(cè)設(shè)備行業(yè)調(diào)查研究
- 統(tǒng)編版一年級(jí)語文下冊(cè)第八單元大單元整體作業(yè)設(shè)計(jì)含答案
- DB21-T 3413-2021地下工程自防護(hù)混凝土結(jié)構(gòu)耐久性技術(shù)規(guī)程
評(píng)論
0/150
提交評(píng)論