CAP 6412 Advanced Computer Vision - CS Departmentbgong/CAP6412/lec2.pdfHuman-centered CV 3D CV...

CAP6412AdvancedComputerVision

Website:http://www.cs.ucf.edu/~bgong/CAP6412.html

Jan14,2016

• Administrivia• Neuralnetworks&backpropagation (PartI)• FundamentalsofConvolutionalNeuralNetworks(CNN),byFareeha

Webcourse vs.Coursehomepage

• Webcourse:https://webcourses.ucf.edu/

• Announcements• CheckyourUCFemail!

• Homeworksubmission

• Coursehomepage:http://www.cs.ucf.edu/~bgong/CAP6412.html

• Alltheothers• Lecturenotes,papers,linkstoresources,syllabus,etc.• Bookmarkandcheckregularly

Topicsyouhavechosen

01234567

TentativescheduleWeek2 CNNvisualization&objectrecognition

Week3 CNN&objectlocalization

Week4 CNN &transferlearning

Week5 CNN &segmentation,super-resolution

Week6 CNN&videos(opticalflow,pose)

Week7 Imagecaptioning&attentionmodel

Week8 Visualquestionanswering

Week9 Attentionmodel,aligningbookswithmovies

Week10--16 Video:tracking,action,surveillanceHuman-centered CV3DCVLow-levelCV,etc.

Nextweek:CNNvisualizatin &objectrecognition

Tuesday(01/19)

[ILSVRC] Russakovsky, Olga, Jia Deng, Hao Su, Jonathan Krause,Sanjeev Satheesh, Sean Ma, Zhiheng Huang et al. “Imagenet largescale visual recognition challenge.” International Journal of ComputerVision (2014): 1-42.[152 layers] He, Kaiming, Xiangyu Zhang, Shaoqing Ren, and Jian Sun.“Deep Residual Learning for Image Recognition.” arXiv preprintarXiv:1512.03385 (2015).

Thursday(01/21)

[Visualization] Zeiler, Matthew D., and Rob Fergus. “Visualizing andunderstanding convolutional networks.” In Computer Vision–ECCV2014, pp. 818-833. Springer International Publishing, 2014.Zhou, Bolei, Aditya Khosla, Agata Lapedriza, Aude Oliva, and AntonioTorralba. “Object detectors emerge in deep scene cnns.” arXivpreprint arXiv:1412.6856 (2014).

Link willbesenttoyourUCFemails

Biologicalneurons

• Humanbrainshasabout10billionnuerons• Eachconnectedto10Kotherneurons• Aneuronfiresifthesumofelectrochemicalinputsexceedssomethreshold

Imagecredit:cs.stanford.edu/people/eroberts

Artificialneurons--- perceptrons

• IntroducedbyRosenblattin1958• Thebasicbuildingblocks for(notall)neuralnetworks

Imagecredit:www.hiit.fi/u/ahonkela/dippa/node41.html

y = '(nX

wixi + b)

= '(wTx+ b)

'(·) : activation function

Popularactivationfunctions

-10 -5 0 5 10-1

1Binary step

-10 -5 0 5 10-1

1Logistic

-10 -5 0 5 10-1

-10 -5 0 5 100

10Rectified Linear Unit (ReLU)

'(x) =

(0 if x < 0

1 if x � 0'(x) =

1 + exp(�x)

'(x) = tanh(x)

exp(x)� exp(�x)

exp(x) + exp(�x)

'(x) =

(0 if x < 0

x if x � 0

Artificialneurons--- perceptrons

• SupportVectorMachines• Logisticregression• AND• OR• NOT• XOR?

• Linearregression

Imagecredit:www.hiit.fi/u/ahonkela/dippa/node41.html

Buildingneuralnetworksfromperceptrons

• NextTuesday

Convolutional Neural

Networks

Fareeha Irfan

Outline ❏ Background❏ Applications: Convnets for object recognition and language❏ How to design convolutional layers❏ How to design pooling layers❏ How to integrate back-propagation in Convnets❏ How to build convnets in torch❏ AlexNet

Background❏ Complex classification tasks❏ Object Recognition in Images:

❏ grayscale: 32 x 32 = 1024 pixels

❏ rbg: 32 x 32 x 3 = 3072 pixels

❏ Fully-connected NN becomes computationally intensive

Algorithm that mimics the brain..

● Neural connections● Neurons activated during learning

Convnet Applications● Image/Object Recognition: Can predict who is in the image, what pose are they

in.● Natural Language Processing: Predict sentiments about sentences to classify

tweets. Extract summaries by finding sentences that are most predictive.● Drug Discovery: Predicting the interaction between molecules and biological

proteins can be used to identify potential treatments.

Some Common Libraries:

● Caffe : Supports both CPU & GPU. Developed in C++● Torch framework: Written in C● Cuda-convnet: Implementation in CUDA

A Simple Neural Network

Activation Functions:

● Sigmoid ● Hyperbolic● Tangent● ReLU (Rectified Linear Unit)

Neural Network

Layer 1Layer 2

Layer 3

Convnet Overview

Neural Network Layer 1 (C1)parameters:

(32*32+1)*(28*28+1)*6= 4827750

ConvNetLayer 1 (C1)parameters:(5*5+1)*6

Convolutional Layery : Output of the convolutionx : Map with K channelsK′ : Total filters, generating a K′ dimensional map y

Back-propagation

Back-propagation for Conv Layer

Pooling Layer

A pooling operator operates on individual feature channels, coalescing nearby feature values into one by the application of a suitable operator.

Common choices include max-pooling (using the max operator) or sum-pooling (using summation).

Max-pooling is defined as:

Pooling Layer

Convnet

● 60 million parameters ● 650,000 neurons● 5 convolutional layers ( followed by

max-pooling layers)● 3 fully-connected layers with ● a 1000-way softmax final layer

Reduces the top-1 error rate by over 1%

TrainingUsing stochastic gradient descent and the backpropagation algorithm (repeated application of the chain rule)

Start with some initialized weights

Optimize so the correct label is predicted

Propagate errors back, and update weights to take a small step in the direction that minimizes the error

http://image-net.org/challenges/LSVRC/2012/supervision.pdf

Stochastic Gradient Descent Learning

CAP 6412 Advanced Computer Vision - CS Departmentbgong/CAP6412/lec2.pdfHuman-centered CV 3D CV...

Documents

Clôtures ALUMINIUM PVC - Menuiserie LACASSAGNE...DINA CV 67 ARRO 2 CV 67 ARRO 3 CV 20 TARRANO CV 85 T PELCO CV 67 T DOLTI CV 14 NOCETA CV 38 APIA CV 15 CASALTA CV 18 NOVALÉ CV 20

CAP 6412 Advanced Computer Vision - CS Departmentbgong/CAP6412/lec10.pdf · CAP 6412 Advanced Computer Vision ... , weight decay (2e-4), #iteration 10,000 ... skip-layer net architecture;

CV Elektriska cirkulära kanalvärmare · m/s Pa Sid 4 | Kap 1 Sortimentsöversikt ELEKTRISKA CIRKULÄRA KANALVÄRMARE Storleksbeteckning CV 10 CV 12 CV 16 CV 20 CV 25 CV 31 CV 40

CAP 6412 Advanced Computer Vision - UCF CRCV · Email --- the best way to reach me • Put [CAP6412] in subject line • Summarize message in subject line • Ex: [CAP6412] Meeting

別売部品 CV 12DA CV 14DBL CV 18DBL CV 350V別売部品コードレスマルチツール（CV 12DA・CV 14DBL・CV 18DBL）/マルチツール（CV 350V）ブレード（STARLOCK）被

CAP 6412 Advanced Computer Visionbgong/CAP6412/lec4.pdf · BMVC 2014. {Major}J. Hosang, R. Benenson, P. Dollár, and B.Schiele.What makes for effective detectionproposals?PAMI 2015

CAP 6412 Advanced Computer Visionbgong/CAP6412/lec5.pdf · CAP 6412 Advanced Computer Vision ... Human-centered CV 3D CV Low-level CV, etc. ... • Object detection proposals, by

Stahovací pásek C V · cv-140sw cv.150sw cv.165W CV-200MW cv-250sw CV-300MW cv-400sw CV-140SKW CV-150SKW CV-165KW CV-200MKW CV-250SKW 142.0 155.0 165.0 203.0 250.0 300.0

COPPA VENET Edizione 2016 O - UISP...SECONDO TURNO - Gare in ordine cronologico CV 47 CV 54 CV 56 CV 53 CV 51 CV 58 CV 48 CV 49 CV 43 Golden Set: San Biagio CV 60 CV 55 Golden Set:

102B3 102A5 302A5 Tarla Parcela cv-42- cv-42- cv-42- cv-42- cv-42- cv-42- cv-42- cv-42- cv-42- Pagina 3 din 5 . Tel. Numele prenumele proprietarului/ proprietarilor Pal Janos Pal

Escond id o C reek - SanDiegoCounty.gov · Cv Cv Cv Cv Cv Cv Cv Cv Cd 2 NOHA Qa 3 Cv 2 Cd 4 Ap 2 Cv 2 Cd 2 Cd 5 Cd 4 Cv 5 Cv 4 WTKI WEBL TUVU Cv 7 Cv 6 Cv 4 Cv 2 Cv 3 TUVU BAOW Cv

DISTRICT 05 PERMANENT SITE MAP · 7 4 7 6 £¤22 £¤22 cv 145 cv 662 cv 873 cv 191 cv 191 cv 419 cv 443 cv 443 cv 191 cv 715 cv 737 cv 534 cv 443 cv 183 cv 183 cv 903 cv 903 cv 73

Série 5E5060E 5070E 5080E 5078E 5090E MOTOR Potência nominal do Motor 60 cv 70 cv 80 cv 78 cv 90 cv Potência nominal da TDP 45 cv 55 cv 65 cv 62 cv 74 cv Modelo John Deere / …

T 3 S - Montana DEQ · CV-R3 CV-1B CV-1 CV-2 CV-3 CV-3A CV-2A CV-1A CV-4 CV-9 CV-10 CV-5 CV-7 EP-3A EP-2A EP-1A Mining Area Mine Permit Boundary Tract Boundary Township Lines Section

FRA-RFM · CV = impregnated cellulose 25 µm β>2 CV CV CV CV CV CV CV CV CV CV CV CV CV CV CV CV CV CV MS = wire mesh 60 µm MS MS MS MS MS MS MS MS MS MS MS MS MS MS MS MS MS MS

CAP 6412 Advanced Computer Vision - CS Departmentbgong/CAP6412/lec1.pdfEmail --- the best way to reach me • Put [CAP6412] in subject line • Summarize message in subject line •

CAP 6412 Advanced Computer Vision - UCF Computer Sciencebgong/CAP6412/lec7.pdf · Week 3 CNN & object localization Week 4 CNN& transfer learning Week 5 CNN& segmentation, super-resolution

Mapa de Tráfico del Área de Carreteras de la … · cv-557 cv-421 cv-422 cv-576 cv-428 cv-506 cv-395 cv-347 cv-310 cv-580 cv-3 96 cv-424 cv-473 cv-580 cv-380 cv-470 cv-435 cv-424

Associative Embedding: End-to-End Learning for Joint ...crcv.ucf.edu/courses/CAP6412/Spring2018/Associative Embedding_ En… · Instance Segmentation. Problem. ... Shoulders, Elbows,

€¦ · cv-43- cv-43- cv-43- cv-43- cv-43- cv-43- cv-43- cv-43- cv-43- cv-43- Nagy L Akos, Nagy F. Csaba Albert, Vajna L. Lajos, Nagy A Csilla 0,45 lldiko,Nagy A. Zita Eniko, Nagy