OpenCV - OpenCV DevZone

1

h1. *Ideas Page for OpenCV Google Summer of Code 2016 (GSoC 2016)*

2

1

3

1

4

1

*OpenCV 3D recognition and pose from last year.*

5

1

!3D_recog.jpg!

6

1

7

1

----

8

1

9

1

h2. General Information:

10

1

11

1

* "GSoC 2016 site: https://developers.google.com/open-source/gsoc/":https://developers.google.com/open-source/gsoc/

12

1

* Mailing list for OpenCV GSOC 2015: [email protected]

13

1

** Group site https://groups.google.com/forum/#!forum/opencv-gsoc-2015

15

1

16

1

** *Next important dates:*

32

1

33

1

*Times:*

34

1

*  UTC to PDT (California uses PST in the winter (from Nov 1st) and PDT in the summer (from March 8)).

35

1

** "utc time converter":http://www.timebie.com/std/gmt.php

36

1

37

1

*Resources:*

38

1

39

1

* "OpenCV Developers Site":http://code.opencv.org/projects/opencv/wiki

40

1

* "OpenCV User's Site":http://opencv.org

41

1

* [[How_to_contribute|How to do a pull request/How to Contribute Code]]

42

1

* Source Code can be found at "GitHub":https://github.com/Itseez/opencv.

43

1

** To get the source code, point your git client to @git://github.com/Itseez/opencv.git@. Here are [[Working_with_OpenCV_git_repository|instructions]].

44

1

<pre>

45

1

git clone git://github.com/Itseez/opencv.git

46

1

</pre>

47

1

** The rest of the history plus matches between git commits and SVN revisions are stored at a separate "OpenCV Attic" repository: git://code.opencv.org/opencv_attic.git.

48

1

** Also OpenCV Extra was put to a separate repository: git://code.opencv.org/opencv_extra.git.

49

1

* [[OpenCV Development Process]]

50

1

** [[Meeting_notes|Developer meeting notes]]

51

1

52

1

h2. How you will be evaluated if you are an accepted student

53

1

54

1

* Student projects to be paid only if:

55

1

** *Midterm:*

56

1

*** You must generate a pull request

57

1

**** That builds

58

1

**** Has at least stubbed out functionality

59

1

**** With appropriate Doxygen documentation

60

1

**** Has at least stubbed out unit test

61

1

**** Has a stubbed out example of use that builds

62

1

** *End of summer:*

63

1

*** A full pull request

64

1

**** Full Doxygen documentation

65

1

**** A good unit test

66

1

**** Example of use code

67

1

*** Create a (short!) Movie (preferably on Youtube, but any movie) that demonstrates your code

68

1

**** We use this to create an overall summary. Past years:

69

1

***** "The 2015 Movie":https://www.youtube.com/watch?v=OUbUFn71S4s

70

1

***** "The 2014 Movie":https://www.youtube.com/watch?v=3f76HCHJJRA

71

1

***** "The 2013 Movie":https://www.youtube.com/watch?v=_TTtN4frMEA

72

1

73

1

74

1

h2. For students interested in applying

75

1

76

1

# You *must* already know how to program well in C++

77

1

** Some projects may require Python or Matlab skills

78

1

# Discuss projects below or other ideas with us https://groups.google.com/forum/?fromgroups#!forum/opencv-gsoc-2015

80

1

# Post the project from below or your own agreed on project on the GSoC site to the Open Source Computer Vision organization

81

1

82

1

h2. For computer vision professionals interested in mentoring

83

1

85

1

# Request to be a mentor for OpenCV there.

86

1

# We accept or reject you

87

1

# Students submit projects and you

88

1

** find a project you want to mentor (several students might try for the same project)

89

1

** contact us about it through [email protected] so we can be sure there is not someone already handling that project

90

1

** contact the students to assess which one if any are capable of coding that up in a summer.

91

1

# We get a slot allocation from Google, the administrators then _"spend"_ the slots in order of priority

92

1

# Students must finally actually accept to do that project (some sign up for multiple organizations and then choose)

93

1

94

1

*If* you are accepted as a mentor *and* you find a suitable student *and* we give you a slot *and* the student signs up for it, +*then*+ you are a mentor.

95

1

** It sounds harder than it is.

96

1

97

1

98

1

<pre>

99

1

\/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/  \/

100

1

</pre>

101

1

102

1

h1. 2016 Project Ideas:

103

1

104

1

*Students may propose their own projects (give us a clear summary and why you can do this project). However, below are some of our priorities for this year* Contact us and/or discuss ideas at https://groups.google.com/forum/?fromgroups#!forum/opencv-gsoc-2015

105

1

106

1

_These are not in order of priority_

107

1

108

1

*1 Google Cardboard*

109

1

* Do visual odometry, SLAM etc for Google Cardboard

110

1

111

1

*2 Projection mapping (Projected Augmented Reality)*

112

1

* Calibrate a camera to a projector

113

1

* Put out structured light to learn the 3D map of what's in front of the camera

114

1

* Project the camera image, suitably rotated and shifted onto the scene

115

1

* Unity or some other program to start transforming the scene for projection mapping

116

1

** See http://www.creativebloq.com/video/projection-mapping-912849 to get ideas of what you can do with this

117

1

118

1

*3 Update OpenCV file storage so that it's readable by Python in OpenCV (move to YAML 1.2 Edition 3)*

119

1

* Mentor Prasanna

120

1

* OpenCV C++ writes out its own YAML tags which makes it incompatible with it's own Python and with external python readers

121

1

* Speedup fs reading/writing, add JSON support, add base64 (+RLE) support for large amounts of numerical data

122

1

* References

123

1

** YAML http://www.yaml.org

124

1

** Problem statement: http://stackoverflow.com/questions/11141336/filestorage-for-opencv-python-api

125

1

126

1

*4 Stereo correspondence improvements*

127

1

* More accurate and faster stereo correspondence (StereoBM/StereoSGBM) by using LBP/BRIEF/DAISY etc. per-pixel descriptors instead of sum-of-absolute-differences.

128

1

129

1

*5 Self-contained video codec(s)*

130

1

* There is draft version of self-contained motion jpeg encoder to be integrated to OpenCV.

131

1

* We will also need decoder, 12-bit jpeg support, multi-threaded encoder and decoder.

132

1

* There are some other codecs, like libvpx, which we can support and link right into highgui, without dependency from ffmpeg/libav.

133

1

134

1

*6 Deep learning*

135

1

* It would be cool if OpenCV could load and run deep networks trained with popular DNN packages like Caffe, Theano or Torch.

136

1

* GPU/OpenCL optimization and/or other optimizations (that can partly be contributed back to the upstream projects) are also encouraged.

137

1

138

1

*7 Better interactive camera calibration*

139

1

* Modern algorithms would allow us to recompute camera parameters after almost each frame, compute reprojection error, compute confidence intervals for the parameters.

140

1

* the partially occluded calibration patterns could be handled somehow to, especially if we want to estimate intrinsic parameters only.

141

1

* the application may show us the areas where the corner borders have been shown so that user can see how good are the data

142

1

143

1

*9 Color Calibration and white balance*

144

1

* With McBeth chart and with gray world algorithm (improve exisiting color constancy algorithms)

145

1

* Possibly collect deep learning way of doing this (needs large training set)

146

1

147

1

*10 Realtime image preprocessing algorithms (real time on mobile)*

148

1

* We and many other OpenCV users would definitely enjoy very fast denoising, color improvements, white balance etc. algorithms.

149

1

150

1

*11 Tracking improvements*

151

1

* TLD algorithm is implemented in OpenCV, but it needs improvements and optimizations.

152

1

* ASM/AAM can also be implemented, as well as the particle filter.

153

1

154

1

155

1

*12 PBM (part-based models) improvements*

156

1

* There is latentsvm package in opencv_contrib that needs to be improved (quality-wise and speed-wise).

157

1

* Adding training, even if it involves execution of external svmperf or other tool, is definitely appreciated too.

158

1

159

1

*13 Improving ICF/ACF + <some>Boost training and detection*

160

1

* In opencv_contrib there is initial version of the training application and detector based on integral channel features and boosting. Further improvements of this part would be really useful.

161

1

162

1

*14 lua+OpenCV*

163

1

* the other language that is now used more often for ML and CV (thanks to Torch) is Lua; there are old bindings for a tiny part of OpenCV; could be useful to extend our autogenerated bindings to Lua.

164

1

165

1

*15 human interaction*

166

1

* gesture recognition

167

1

* human skeleton based on 2d or 3d

168

1

169

1

*17 Quad Rotor Contrib*

170

1

* Apply much of the above into a quad rotor drone automatic tracking, mapping navigation etc module

171

1

172

1

173

1

New 2016 ideas:

174

1

175

1

*Architecture

176

1

* Exceptions in OpenCV need to be easily disabled ? (need to check if that's the case or not)

177

1

* Minimal OpenCV: give the functions you want, it gives you a minimal version of the library.

178

1

* reduce the size of OpenCV on Anddroid

179

1

* create a more C++ type (to be able to deal with constness and separate the pointer from the type)

180

1

* create an image type (that would also have color information, maybe calibration too) and implement a proper resize algorithm for that type

181

1

182

1

*Tracking

183

1

* David Tanâs tracking

184

1

2d: https://www.youtube.com/watch?v=W689ffCMDF4

185

1

3d: https://www.youtube.com/watch?v=TqyFEZzhPbE

186

1

187

1

*misc

188

1

* convexHull is implemented with Sklansky's second, 1982 algorithm, which is incorrect in some case, and suboptimal complexity-wise. cf. http://cgm.cs.mcgill.ca/~athens/cs601/.

189

1

cv::contourArea(cv::convexHull(contour)) for a convex contour is different from cv::contourArea(contour), somtimes by nearly 30%.

190

191

192

193

<pre>

194

1

/\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\ /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\

195

1

</pre>

196

197

199

201

202

203

*GSoC Progress. Need Pull requests (don't have to be complete) to pass by July 3rd*

204

1

205

1

*Mailing Lists (only for use of Mentors and Students, please to not request to join these lists otherwise):*

206

1

* Mentors

207

1

** opencv-gsoc2015-mentors

208

1

* Students

209

1

** opencv-gsoc2015-students

210

1

211

1

* [[How_to_contribute|How to do a pull request]]

212

1

213

1

*Pre-Midterm status:*

214

1

* *Pull Request In* Antonella Cascitelli - Vladimir Tyan ([email protected]): TLD improvements - https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/auron/5629499534213120.

215

1

* *Pull Request In* Bo Li - Baisheng Lai ([email protected]). Omnidirectional Cameras Calibration and Stereo 3D Reconstruction. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/baisheng/5629499534213120

216

1

* _*No Pull Request*_ Vincent Rabaud - Edgar Riba ([email protected]). Structure From Motion module. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/edgarriba/5693417237512192

217

1

* *Pull Request In* Bence Magyar - Jiaolong Xu ([email protected]). Improving PBM in OpenCV. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/jiaolong/5718998062727168

218

1

* *Pull Request In* Fernando J. Iglesias Garcia - Laksono Kurnianggoro ([email protected]). Real-time Multi-object Tracking using Kernelized Correlation Filter. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/kurnianggoro/5629499534213120

219

1

* _*No Pull Request*_, but previously Vadim Pisarevsky - Lluis Gomez ([email protected]). Improve and expand Scene Text Detection module in OpenCV. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/lluisgomez/5751725713522688

220

1

* _*No Pull Request*_ Michael Harville, Gary Bradski - Naoto Hieda ([email protected]). Projection mapping (Projected Augmented Reality). https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/micuat/5707702298738688

221

1

* *Pull Request In* [email protected] - Mircea Paul Muresan ([email protected]). Stereo correspondence improvements. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/mmp131316/5741031244955648[Andrey can not mentor the student. Does anyone want to take one?]

222

1

* *Pull Request In* Delia, Stefano Fabri, Claudia Rapuano - RobertaRavanelli ([email protected]). Structured-Light System Calibration for OpenCV. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/roberta_ravanelli/5676830073815040

223

1

* "done" Pull Maksim Shabunin - Alexander Bokov ([email protected]). Recent advances in edge-aware filtering. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/sbokov/5629499534213120

224

1

* _*No Pull Request*_ Prasanna, Gary Bradski - Sergio Garrido ([email protected]). Charuco board: Chessboard+ArUco for camera calibration. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/sgarrido/5629499534213120

225

1

* _*No Pull Request*_ Alexander Bovyrin - Vlad Shakhuro ([email protected]). ICF detector improvements. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/shahurik/5717271485874176

226

1

* *Pull Request In* Vadim Pisarevsky - Seon-Wook Park ([email protected]). Real-time image preprocessing on ARM. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/swook/5798429187899392

227

1

* *Pull Request In* Anatoly Baksheev, Douglas Lee - Vitaliy Lyudvichenko ([email protected]). Implementation of universal interface for deep neural network frameworks. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/vludv/5657382461898752

228

1

* *Pull Request In* Manuele Tamburrano, Stefano Fabri - Yida Wang ([email protected]). Deep Learning & GPU Implementation Proposal. https://www.google-melange.com/gsoc/proposal/review/org/google/gsoc2015/wangyida/5629499534213120

229

1

230

1

Project notes will be compiled "here":https://sites.google.com/a/opencv.org/opencv/gsoc-2015-project-notes .

231

232

233

234

h2. Mentors

235

1

236

1

*Actual signed up mentors*

237

1

238

1

* Alexander Mordvintsev	zzznah(+)gmail.com

239

1

* Sid Bao	ybao(+)magicleap.com

240

1

* Vincent Rabaud	vincent.rabaud(+)gmail.com

241

1

* Vadim Pisarevsky	vadim.pisarevsky(+)gmail.com

242

1

* Adrian Kaehler	therealadrian(+)gmail.com

243

1

* Terry Boult	tboult(+)vast.uccs.edu

244

1

* spmallick	spmallick(+)taaz.com

245

1

* Serge Belongie	sjb344(+)cornell.edu

246

1

* Stefano	s.fabri10(+)gmail.com

247

1

* Prasanna	pras.bits(+)gmail.com

248

1

* Pablo Alcantarilla	pablofdezalc(+)gmail.com

249

1

* Bence Magyar	mw.mzperx(+)gmail.com

250

1

* Manuele	manuele.tamburrano(+)gmail.com

251

1

* Grace Vesom	grace.vesom(+)gmail.com

252

1

* Open Source Computer Vision Library (OpenCV)	garybradski(+)gmail.com

253

1

* Douglas Lee	dougabug(+)gmail.com

254

1

* Claudia Rapuano	c.rapuano(+)gmail.com

255

1

* Antonella Cascitelli	antonellacascitelli(+)gmail.com

256

1

* Anatoly Baksheev	anatoly.baksheev(+)itseez.com

257

1

* Alexander	alexander.shishkov(+)itseez.com

258

1

* Alexander Smorkalov	alexander.smorkalov(+)itseez.com

259

1

* Alexander Bovyrin	alexander.bovyrin(+)itseez.com

260

261

262

263

In the below, get rid of the ''-delete-'' to make the emails work.

264

1

265

1

 Anatoly Baksheev

266

1

 Researcher, Vision Algorithms on GPU

267

1

 Argus/Itseez

268

1

 [email protected]

269

1

270

1

  Alexander Bovyrin

271

1

  PhD, Senior Researcher

272

1

  Argus/Itseez founder

273

1

  NNU Lecturer

274

1

  [email protected]

275

1

276

1

 Gary Bradski

277

1

 Founder, Industrial Perception Inc.

278

1

 Consulting Prof. Stanford U.

279

1

 OpenCV Founder, Technical Content Owner, GSoC Admin

280

1

 Co-author of Learning OpenCV Book

281

1

 http://www.willowgarage.com/pages/people/gary-bradski-senior-researcher

282

1

 [email protected]

283

1

284

1

 Antonella Cascitelli

285

1

 Grad student, University of Rome

286

1

 [email protected]

287

1

288

1

 Eric Christiansen

289

1

 PhD, UCSD

290

1

 [email protected]

291

1

292

1

 Stefano Fabri

293

1

 CRR Team leader, University of Rome

294

1

 [email protected]

295

1

296

1

 Victor Eruhimov

297

1

 OpenCV founding team/Senior Researcher

298

1

 Argus/Itseez founder

299

1

 NNU Lecturer

300

1

 [email protected]

301

1

302

1

 Adrian Kaehler

303

1

 Principle Engineer, Applied Minds

304

1

 Co-author of Learning OpenCV Book.

305

1

 [email protected]

306

1

307

1

 Peter Karasev

308

1

 PhD Student, MINERVA Research Group

309

1

 Georgia Tech

310

1

 [email protected]

311

1

312

1

 Vadim Pisarevsky

313

1

 OpenCV founding team/Czar

314

1

 [email protected]

315

1

316

1

 Manuele Tamburrano

317

1

 Grad Student, University of Rome

318

1

 [email protected]

319

1

320

1

 Vincent Rabaud

321

1

 Principle Engineer, Aldebaran Robotics

322

1

 [email protected]

323

1

324

1

  Claudia Rapuano

325

1

  Grad student, University of Rome

326

1

  [email protected]

327

1

328

1

 Grace Vesom

329

1

 Senior Software Engineer, Magic Leap

330

1

 -delete-grace.vesom@-delete_gmail.com

331

1

332

1

 Bence Magyar

333

1

 Pal Robotics

334

1

 [email protected]

335

1

336

1

 Pablo Alcantarilla

337

1

 Toshiba Research Europe Ltd.

338

1

 [email protected]

339

1

340

1

h3.  Back up Mentors

341

1

342

1

 Mark Asbach

343

1

 Fraunhofer IAIS

344

1

 Schloss Birlinghoven

345

1

 Sankt Augustin, Germany

346

1

 http://mmprec.iais.fraunhofer.de/asbach.html

347

1

 [email protected]

348

1

349

1

350

1

 Nicolas Saunier, Ph.D.

351

1

 Assistant Professor

352

1

 Civil, Geological and Mining Department (CGM)

353

1

 Ãcole Polytechnique de MontrÃ©al

354

1

 http://nicolas.saunier.confins.net

355

1

 [email protected]

356

1

357

1

358

1

 Alexander Mordvintsev

359

1

 Software Engineer

360

1

 http://znah.net

361

1

 [email protected]

362

1

363

1

 Andrey Morozov

364

1

 Software Engineer

365

1

 Argus/Itseez

366

1

 [email protected]

367

1

369

1

370

1

"OpenCV GSoC Application":http://code.opencv.org/projects/gsoc2013/wiki/GSoC2013OpenCVApplication

371

1

372

1

h2. Previous Years

373

1

374

1

* 2015 "GSoC 2015 Ideas Page":http://code.opencv.org/projects/opencv/wiki/GSoC_2015

375

1

** "The 2015 Movie":https://www.youtube.com/watch?v=OUbUFn71S4s

376

1

* 2014 "GSoC 2014 Ideas Page":http://code.opencv.org/projects/opencv/wiki/GSoC_2014

377

1

** "The 2014 Movie":https://www.youtube.com/watch?v=3f76HCHJJRA

378

1

* 2013 "GSoC 2013 Ideas Page":http://code.opencv.org/projects/gsoc2013/wiki

379

1

** "The 2013 Movie":https://www.youtube.com/watch?v=_TTtN4frMEA

380

1

* 2012 Ideas site http://code.opencv.org/projects/gsoc2012/wiki/Wiki/

381

1

* 2011 Ideas site http://opencv.willowgarage.com/wiki/GSOC_OpenCV2011

382

1

** 2011 Application GSoC page is http://opencv.willowgarage.com/wiki/GSoC2011OpenCVApplication

383

1

* 2010 Ideas site http://opencv.willowgarage.com/wiki/GSOC_OpenCV2010

384

1

** 2010 Application page http://opencv.willowgarage.com/wiki/ApplicationGSOC2010

385

1

386

1

----

Login	Password