-
Notifications
You must be signed in to change notification settings - Fork 0
/
past-research.html
1310 lines (1077 loc) · 85.3 KB
/
past-research.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=UA-176270417-1"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'UA-176270417-1');
</script>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<meta http-equiv="Content-Style-Type" content="text/css">
<meta name="generator" content="pandoc">
<title>Ravi Kiran Sarvadevabhatla</title>
<script async="" src="./research_files/analytics.js"></script><script src="./research_files/jquery-1.js"></script>
<link rel="stylesheet" href="./ravikiran_files/bootstrap.css">
<link rel="stylesheet" href="./ravikiran_files/bootstrap-theme.css">
<script src="./ravikiran_files/bootstrap.js"></script>
<style>
/* http://stackoverflow.com/questions/18325779/bootstrap-3-collapse-show-state-with-chevron-icon */
.panel-heading .accordion-toggle:before {
font-family: 'Glyphicons Halflings';
content: "\e114";
float: left;
color: grey;
padding-right: 6px;
}
.panel-heading .accordion-toggle.collapsed:before {
content: "\e080";
}
table, th, td {
border: 0px solid black;
border-collapse: collapse;
}
img:hover{
transition-duration: 0.0s;
transition-timing-function: linear;
opacity:0.5;
}
</style>
<style type="text/css">
</style>
</head>
<body>
<nav class="navbar navbar-default navbar-static-top" role="navigation">
<div class="container">
<ul class="nav navbar-nav">
<li><a href="index.html"><font style="font-size:20px; font-weight:500;" color="#003380">Ravi Kiran Sarvadevabhatla</font></a></li>
<li><a href="publications.html"><font size="4px" color="#E62E00"><b>Publications</b></font></a></li>
<li><a href="research.html"><font size="4px" color="#E62E00"><b>Current Research</b></font></a></li>
<li><a href="past-research.html"><font size="4px" color="#E62E00"><b>Past Research</b></font></a></li>
</ul>
</div>
</nav>
<div class="container">
<a name="Research"></a>
<p style="margin:-15px 0px 0px 0px;"></p>
<div class="page-header"><h2 id="research">Past Research</h2></div>
<div class="row">
<div class="pull-left" style="width:1.4%;">
</div>
<div class="pull-left" style="width:29%;">
<div class="box" style="width:100%;">
<div align="center">
<a data-toggle="collapse" data-parent="#detection" href="research.html#detection-info"><img width="100%" src="./research_files/dets_med1_small.png"></a>
<br>
</div>
</div>
<p style="margin:5px 0px 0px 0px;"></p>
<font style="font-size:18px; font-weight:500;">Recognition and Representation</font><a data-toggle="collapse" data-parent="#detection" href="research.html#detection-info"><img style="height:26px; margin-top:0px;" align="right" src="./research_files/info.png"></a><br>
</div>
<div class="pull-left" style="width:4%;">
</div>
<div class="pull-left" style="width:29%;">
<div class="box" style="width:100%;">
<div align="center">
<a data-toggle="collapse" data-parent="#moviebook" href="research.html#moviebook-info">
<img width="100%" src="./research_files/mb1.jpg">
</a>
<br>
</div>
</div>
<p style="margin:5px 0px 0px 0px;"></p>
<font style="font-size:18px; font-weight:500;">Visualization</font><a data-toggle="collapse" data-parent="#moviebook" href="research.html#moviebook-info"><img style="height:26px; margin-top:0px;" align="right" src="./research_files/info.png"></a><br>
</div>
<div class="pull-left" style="width:4%;">
</div>
<div class="pull-left" style="width:29%;">
<div class="box" style="width:100%;">
<div align="center">
<a data-toggle="collapse" data-parent="#scene" href="research.html#scene-info"><img width="100%" src="./research_files/cvpr15aptlayout_med.png"></a>
<br>
</div>
</div>
<p style="margin:5px 0px 0px 0px;"></p>
<font style="font-size:18px; font-weight:500;">Affective Computing</font><a data-toggle="collapse" data-parent="#scene" href="research.html#scene-info"><img style="height:26px; margin-top:0px;" align="right" src="./research_files/info.png"></a><br>
</div>
<br>
</div>
<p style="margin:30px 0px 0px 0px;"></p>
<div class="row">
<div class="pull-left" style="width:1.4%;">
</div>
<div class="pull-left" style="width:29%;">
<div class="box" style="width:100%;">
<div align="center">
<a data-toggle="collapse" data-parent="#imagetext" href="research.html#imagetext-info"><img width="100%" src="./research_files/text1_small.png"></a>
<br>
</div>
</div>
<p style="margin:5px 0px 0px 0px;"></p>
<font style="font-size:18px; font-weight:500;">Human-Robot Interaction</font><a data-toggle="collapse" data-parent="#imagetext" href="research.html#imagetext-info"><img style="height:26px; margin-top:0px;" align="right" src="./research_files/info.png"></a><br>
</div>
<div class="pull-left" style="width:4%;">
</div>
<div class="pull-left" style="width:29%;">
<div class="box" style="width:100%;">
<div align="center">
<a data-toggle="collapse" data-parent="#segmentation" href="research.html#segmentation-info"><img width="100%" src="./research_files/SketchParse.png"></a>
<br>
</div>
</div>
<p style="margin:5px 0px 0px 0px;"></p>
<font style="font-size:18px; font-weight:500;">Semantic Object Segmentation</font><a data-toggle="collapse" data-parent="#segmentation" href="research.html#segmentation-info"><img style="height:26px; margin-top:0px;" align="right" src="./research_files/info.png"></a><br>
</div>
<div class="pull-left" style="width:4%;">
</div>
<div class="pull-left" style="width:29%;">
<div class="box" style="width:100%;">
<div align="center">
<a data-toggle="collapse" data-parent="#cognitivemodel" href="research.html#cognitivemodel-info"><img width="100%" src="./research_files/cognitive-models.png"></a>
<br>
</div>
</div>
<p style="margin:5px 0px 0px 0px;"></p>
<font style="font-size:18px; font-weight:500;">Cognitive Modelling</font><a data-toggle="collapse" data-parent="#cognitivemodel" href="research.html#cognitivemodel-info"><img style="height:26px; margin-top:0px;" align="right" src="./research_files/info.png"></a><br>
</div>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<div id="detection-info" class="panel-collapse collapse out" style="padding:0% 3% 1% 3%; border-radius:6px; border-style: solid; border-color: #666666;">
<p style="margin:10px 0px 0px 0px;"></p>
<font style="font-size:1.34em; font-weight:500;">Recognition and Representation</font>
<div class="media">
<p align="justify">Visual Recognition encompasses many important sub-topics in Computer Vision. The recognition problem and related approach are intimately tied to the specific task. In the course of my work, I have explored recognition of objects for various modalities (hand-drawn sketches, artistic depictions, photos), printed text characters from Indian language documents and human activities. In particular, my work on online sketch recognition is currently state-of-the-art for large-scale sketch object recognition.</p>
<p align="justify">Object depictions tend to be available across a spectrum which is characterized by level and richness of detail. At one extreme, we have color photographs and at the other extreme, we have black-and-white line drawings. In my Ph.D. work, I posed the question: Starting from an object line-drawing, can we "extend" the spectrum mentioned previously and create an even sparser yet recognizable representation of the object ? In the course of my work, I have shown that we can answer this question in the affirmative. Specifically, I designed a novel sketch-based visual representation for objects called <i>category-epitome</i>. My work also includes extensive analysis of <i>category-epitomes</i> across a large number of object categories and reveals many interesting aspects about this novel representation.
<p style="margin:30px 0px 0px 0px;"></p>
<h3> Relevant Publications</h3>
<br>
<p style="margin:-10px 0px 0px 0px;"></p>
<font style="font-size:1.11em; font-weight:500;">Recognition</font>
<br>
<br>
<p style="margin:-12px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;border:1px solid">
<div class="media">
<div class="pull-left text-center" style="padding:14px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/overview-accv2016-semantic.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">'Part'-ly first among equals: Part-based benchmarking for state-of-the-art object recognition architectures</h4>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px"><strong>Ravi Kiran Sarvadevabhatla</strong>, Shanthakumar Venkatraman, R. Venkatesh Babu</p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>Asian Conference on Computer Vision (<strong>ACCV</strong>)</em>, Taipei,Taiwan ROC, 2016</p>
<!--p style="font-size:13.4px">Arxiv preprint arXiv:1611.03382</p-->
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;">Top-1/Top-5 error based benchmarking results for large-scale object recognition datasets do not reveal which aspects of recognition problem (robustness to occlusion, loss of global detail) the classifiers are good at. Moreover, the overall approach provides a falsely optimistic picture due to dataset bias. In this paper, we propose a novel semantic-part based dataset and benchmarking approach which overcomes the shortcomings mentioned above.</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<a href="https://arxiv.org/pdf/1611.07703v2.pdf" target="_blank" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#PBBMabs" href="#PBBMabs-list">Abstract</a>
<a href="http://val.serc.iisc.ernet.in/pbbm/" target="_blank" class="buttonPP">Project page</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#PBBM" href="#PBBM-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="PBBM-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@article{DBLP:journals/corr/Sarvadevabhatla16b,<br>
author = {Ravi Kiran Sarvadevabhatla and Shanthakumar Venkatraman and R. Venkatesh Babu},<br>
title = {'Part'ly first among equals: Semantic part-based benchmarking for state-of-the-art object recognition systems},<br>
journal = {CoRR}, <br>
volume = {abs/1611.07703}, <br>
year = {2016}, <br>
url = {http://arxiv.org/abs/1611.07703},<br>
timestamp = {Thu, 01 Dec 2016 19:32:08 +0100}, <br>
biburl = {http://dblp.uni-trier.de/rec/bib/journals/corr/Sarvadevabhatla16b}, <br>
bibsource = {dblp computer science bibliography, http://dblp.org} <br> }
</p>
</div>
<div id="PBBMabs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
An examination of object recognition challenge leaderboards (ILSVRC, PASCAL-VOC) reveals that the top-performing classifiers typically exhibit small differences amongst themselves in terms of error rate/mAP. To better differentiate the top performers, additional criteria are required. Moreover, the (test) images, on which the performance scores are based, predominantly contain fully visible objects. Therefore, `harder' test images, mimicking the challenging conditions (e.g. occlusion) in which humans routinely recognize objects, need to be utilized for benchmarking. To address the concerns mentioned above, we make two contributions. <i>First</i>, we systematically vary the level of local object-part content, global detail and spatial context in images from PASCAL VOC 2010 to create a new benchmarking dataset dubbed PPSS-12. <i>Second</i>, we propose an object-part based benchmarking procedure which quantifies classifiers' robustness to a range of visibility and contextual settings. The benchmarking procedure relies on a semantic similarity measure that naturally addresses potential semantic granularity differences between the category labels in training and test datasets, thus eliminating manual mapping. We use our procedure on the PPSS-12 dataset to benchmark top-performing classifiers trained on the ILSVRC-2012 dataset. Our results show that the proposed benchmarking procedure enables additional differentiation among state-of-the-art object classifiers in terms of their ability to handle missing content and insufficient object detail. Given this capability for additional differentiation, our approach can potentially supplement existing benchmarking procedures used in object recognition challenge leaderboards.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/new_burst.png"></div>
</li>
</ul>
<p style="margin:-8px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;border:1px solid">
<div class="media">
<div class="pull-left text-center" style="padding:9px 1.5% 0 0; width:14.1%;" href=""><img width="100%" height="100%" src="./ravikiran_files/Alexnet-GRU-Overview.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Enabling my robot to play Pictionary: Recurrent Neural Networks for sketch recognition</h4>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px"><strong>Ravi Kiran Sarvadevabhatla</strong>, Jogendra Nath Kundu, R. Venkatesh Babu</p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>ACM Multimedia Conference (<strong>ACMMM</strong>)</em>, Amsterdam, The Netherlands 2016</p>
<!--p style="font-size:13.4px">Arxiv preprint arXiv:1608.07711</p-->
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;">We propose a Recurrent Neural Network architecture which exploits the long-term sequential and structural regularities in sketch stroke data for large-scale recognition of hand-drawn object sketches.</p>
<a href="https://arxiv.org/pdf/1608.03369v1.pdf" target="_blank" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#SketchRNNMM16abs" href="#SketchRNNMM16abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#SketchRNNMM16" href="#SketchRNNMM16-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="SketchRNNMM16-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Sarvadevabhatla:2016:EMR:2964284.2967220,<br>
author = {Sarvadevabhatla, Ravi Kiran and Kundu, Jogendra and R, Venkatesh Babu},<br>
title = {Enabling My Robot To Play Pictionary: Recurrent Neural Networks For Sketch Recognition},<br>
booktitle = {Proceedings of the 2016 ACM Conference on Multimedia},<br>
year = {2016},<br>
location = {Amsterdam, The Netherlands}, <br>
pages = {247--251}, <br>
url = {http://doi.acm.org/10.1145/2964284.2967220}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA},<br>
}
</p>
</div>
<div id="SketchRNNMM16abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
Freehand sketching is an inherently sequential process. Yet, most approaches for hand-drawn sketch recognition either ignore this sequential aspect or exploit it in an ad-hoc manner. In our work, we propose a recurrent neural network architecture for sketch object recognition which exploits the long-term sequential and structural regularities in stroke data in a scalable manner. Specifically, we introduce a Gated Recurrent Unit based framework which leverages deep sketch features and weighted per-timestep loss to achieve state-of-the-art results on a large database of freehand object sketches across a large number of object categories. The inherently online nature of our framework is especially suited for on-the-fly recognition of objects as they are being drawn. Thus, our framework can enable interesting applications such as camera-equipped robots playing the popular party game Pictionary with human players and generating sparsified yet recognizable sketches of objects.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
</li>
</ul>
<p style="margin:-8px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;border:1px solid">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/switchnet_full.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">SwiDeN: Convolutional Neural Networks for Depiction Invariant Object Recognition</h4>
<p style="margin:-9.0px 0px 0px 0px;"><strong>Ravi Kiran Sarvadevabhatla</strong>, Shiv Surya, Srinivas SS Kruthiventi, R. Venkatesh Babu</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>ACM Multimedia Conference (<strong>ACMMM</strong>)</em>, Amsterdam, The Netherlands 2016</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;"> In this paper, we present SwiDeN: our Convolutional Neural Network (CNN) architecture which recognizes objects regardless of how they are visually depicted (line drawing, realistic shaded drawing, photograph etc.) </p>
<a href="https://arxiv.org/pdf/1607.08764v1.pdf" class="buttonTT">Paper</a>
<a href="https://github.com/val-iisc/swiden" target="_blank" class="buttonPP">Code</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#SwidenMM16abs" href="#SwidenMM16abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#SwidenMM16" href="#SwidenMM16-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="SwidenMM16-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Sarvadevabhatla:2016:SCN:2964284.2967208,<br>
author = {Sarvadevabhatla, Ravi Kiran and Surya, Shiv and Kruthiventi, Srinivas S S and R., Venkatesh Babu}, <br>
title = {SwiDeN: Convolutional Neural Networks For Depiction Invariant Object Recognition}, <br>
booktitle = {Proceedings of the 2016 ACM on Multimedia Conference}, <br>
series = {MM '16}, <br>
year = {2016}, <br>
isbn = {978-1-4503-3603-1}, <br>
location = {Amsterdam, The Netherlands}, <br>
pages = {187--191}, <br>
numpages = {5}, <br>
url = {http://doi.acm.org/10.1145/2964284.2967208}, <br>
doi = {10.1145/2964284.2967208}, <br>
acmid = {2967208}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
keywords = {convolutional neural networks, deep learning, depiction-invariance, object category recognition}, <br>
}
</p>
</div>
<div id="SwidenMM16abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
Current state of the art object recognition architectures achieve impressive performance but are typically specialized for a single depictive style (e.g. photos only, sketches only). In this paper, we present SwiDeN: our Convolutional Neural Network (CNN) architecture which recognizes objects regardless of how they are visually depicted (line drawing, realistic shaded drawing, photograph etc.). In SwiDeN, we utilize a novel `deep' depictive style-based switching mechanism which appropriately addresses the depiction-specific and depiction-invariant aspects of the problem. We compare SwiDeN with alternative architectures and prior work on a 50-category Photo-Art dataset containing objects depicted in multiple styles. Experimental results show that SwiDeN outperforms other approaches for the depiction-invariant object recognition problem.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
</li>
</ul>
<p style="margin:-8px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;border:1px solid">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/activityrec.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Recognizing Human Activities from Constituent Actions</h4>
<p style="margin:-9.0px 0px 0px 0px;"><strong>Ravi Kiran Sarvadevabhatla</strong>, Karteek Alahari, C.V. Jawahar </p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>National Conference on Communications (<strong>NCC</strong>)</em>, Kharagpur, India 2005</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<a href="http://ieeexplore.ieee.org/document/1227699/" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#Activity05abs" href="#Activity05abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#Activity05" href="#Activity05-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="Activity05-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{activity05,<br>
title={Recognizing Human Activities from Constituent Actions}, <br>
author={Ravi Kiran Sarvadevabhatla, Karteek Alahari, C.V. Jawahar}, <br>
booktitle={National Conference on Communications}, <br>
pages={351--355}, <br>
year={2005},<br>
organization={IEEE} <br>
</p>
</div>
<div id="Activity05abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
Many of the human activities such as Jumping, Squatting
have a correlated spatiotemporal structure. They are composed
of homogeneous units. These units, which we refer
to as actions, are often common to more than one activity.
Therefore, it is essential to have a representation which can
capture these activities effectively. To develop this, we model
the frames of activities as a mixture model of actions and employ
a probabilistic approach to learn their low-dimensional
representation. We present recognition results on seven activities
performed by various individuals. The results demonstrate
the versatility and the ability of the model to capture
the ensemble of human activities.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/Oral-session-icon.png"></div>
</li>
</ul>
<p style="margin:-8px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;border:1px solid">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/ocr.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">A Bilingual OCR for Hindi-Telugu Documents and its Applications</h4>
<p style="margin:-9.0px 0px 0px 0px;">C.V. Jawahar, MNSSK Pavan Kumar, <strong>Ravi Kiran Sarvadevabhatla</strong></p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>Seventh International Conference on Document Analysis and Recognition ( <strong>ICDAR</strong> )</em>, Edinburgh, UK 2003</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.62.228&rep=rep1&type=pdf" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#Ocr03abs" href="#Ocr03abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#Ocr03" href="#Ocr03-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="Ocr03-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Jawahar:2003:BOH:938979.939243, <br>
author = {Jawahar, C. V. and Kumar, M. N. S. S. K. Pavan and Kiran, S. S. Ravi Kiran} , <br>
title = {A Bilingual OCR for Hindi-Telugu Documents and Its Applications}, <br>
booktitle = {Proceedings of the Seventh International Conference on Document Analysis and Recognition - Volume 1}, <br>
series = {ICDAR '03}, <br>
year = {2003}, <br>
isbn = {0-7695-1960-1}, <br>
pages = {408--}, <br>
url = {http://dl.acm.org/citation.cfm?id=938979.939243}, <br>
acmid = {939243}, <br>
publisher = {IEEE Computer Society}, <br>
address = {Washington, DC, USA}, <br>
}
</p>
</div>
<div id="Ocr03abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
This paper describes the character recognition processfrom printed documents containing Hindi and Telugu text. Hindi and Telugu are among the most popular languages in India. The bilingual recognizer is based on Principal Component Analysis followed by support vector classification. This attains an overall accuracy of approximately 96.7%. Extensive experimentation is carried out on an independent test set of approximately 200000 characters. Applications based on this OCR are sketched.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/Oral-session-icon.png"></div>
</li>
</ul>
<p style="margin:-10px 0px 0px 0px;"></p>
<font style="font-size:1.11em; font-weight:500;">Representation</font>
<br>
<br>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;border:1px solid">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/primal-sketches-examples.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Eye of the Dragon : Exploring discriminatively minimalist sketch-based abstractions for object categories</h4>
<p style="margin:-9.0px 0px 0px 0px;"><strong>Ravi Kiran Sarvadevabhatla</strong>, R. Venkatesh Babu</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>ACM Multimedia Conference (<strong>ACMMM</strong>)</em>, Brisbane, Australia 2015</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;"> In this paper, we introduce a novel visual representation derived from freehand sketches of objects. This representation, called <i>category-epitome</i>, is designed to be a sparsified yet recognizable version of the original sketch. We examine various interesting properties of <i>category-epitome</i>s. </p>
<a href="http://dl.acm.org/ft_gateway.cfm?id=2806230&ftid=1633068&dwn=1&CFID=874283070&CFTOKEN=86936959" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#CatEpitomes15abs" href="#CatEpitomes15abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#CatEpitomes15" href="#CatEpitomes15-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="CatEpitomes15-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Sarvadevabhatla:2015:EDE:2733373.2806230,<br>
author = {Sarvadevabhatla, Ravi Kiran and R, Venkatesh Babu}, <br>
title = {Eye of the Dragon: Exploring Discriminatively Minimalist Sketch-based Abstractions for Object Categories}, <br>
booktitle = {Proceedings of the 23rd ACM International Conference on Multimedia}, <br>
series = {MM '15}, <br>
year = {2015},<br>
isbn = {978-1-4503-3459-4}, <br>
location = {Brisbane, Australia}, <br>
pages = {271--280}, <br>
numpages = {10}, <br>
url = {http://doi.acm.org/10.1145/2733373.2806230}, <br>
doi = {10.1145/2733373.2806230}, <br>
acmid = {2806230}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
keywords = {deep learning, freehand sketch, object category recognition}, <br>
}
</p>
</div>
<div id="CatEpitomes15abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
As a form of visual representation, freehand line sketches are typically studied as an end product of the sketching process. However, from a recognition point of view, one can also study various orderings and properties of the primitive strokes that compose the sketch. Studying sketches in this manner has enabled us to create novel sparse yet discriminative sketch-based representations for object categories which we term category-epitomes. Concurrently, the epitome construction provides a natural measure for quantifying the sparseness underlying the original sketch, which we term epitome-score. We analyze category-epitomes and epitome-scores for hand-drawn sketches from a sketch dataset of 160 object categories commonly encountered in daily life. Our analysis provides a novel viewpoint for examining the complexity of representation for visual object categories.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/Oral-session-icon.png"></div>
</li>
</ul>
<a data-toggle="collapse" data-parent="#moviebook" href="research.html#detection-info">close window</a>
</div>
</div>
<br><br>
</div>
<p style="margin:-10px 0px 0px 0px;"></p>
<div id="moviebook-info" class="panel-collapse collapse out" style="padding:0% 3% 1% 3%; border-radius:6px; border-style: solid; border-color: #666666;">
<p style="margin:10px 0px 0px 0px;"></p>
<font style="font-size:1.34em; font-weight:500;">Visualization</font>
<div class="media">
<p align="justify">A proper choice of visualization technique can provide a better and richer understanding of the object category landscape, either for raw visual data or its associated features. A natural way to describe objects is in terms of their named parts (e.g. a bicycle consists of two wheels, a frame and a handlebar set atop the frame in the front). I have used this property to characterize sparsified object representations in terms of relative presence of typical parts associated with the category. My work has also resulted in novel ranking-based method to visualize the intra/inter-category relationships of object feature embeddings.
<p style="margin:30px 0px 0px 0px;"></p>
<h3> Relevant Publications</h3>
<br>
<p style="margin:-12px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:9px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/epivis-epitome-hac.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Analyzing object categories via novel category-ranking measures defined on visual feature emebddings</h4>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px"><strong>Ravi Kiran Sarvadevabhatla</strong>, Raviteja Meesala, Manjunath Hegde, R. Venkatesh Babu</p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>Indian Conference on Graphics, Vision and Image Processing (<strong>ICVGIP</strong>)</em>, Guwahati,India, 2016</p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;">Visualizing category-level features using color-coding is impractical when number of categories is large. This paper presents an approach which utilizes the geometrical attributes of per-category feature collections to order the categories. Our approach enables a novel viewpoint for exploring large-scale object category collections.</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<a href="http://dl.acm.org/ft_gateway.cfm?id=3010037&ftid=1822926&dwn=1&CFID=708922244&CFTOKEN=15898682" target="_blank" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#FeatureViz_icvgipabs" href="#FeatureViz_icvgipabs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#FeatureViz_icvgip" href="#FeatureViz_icvgip-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="FeatureViz_icvgip-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">@inproceedings{Sarvadevabhatla:2016:AOC:3009977.3010037, <br>
author = {Sarvadevabhatla, Ravi Kiran and Meesala, Raviteja and Hegde, Manjunath and R., Venkatesh Babu}, <br>
title = {Analyzing Object Categories via Novel Category Ranking Measures Defined on Visual Feature Embeddings}, <br>
booktitle = {Proceedings of the Tenth Indian Conference on Computer Vision, Graphics and Image Processing}, <br>
series = {ICVGIP '16}, <br>
year = {2016}, <br>
isbn = {978-1-4503-4753-2}, <br>
location = {Guwahati, Assam, India}, <br>
pages = {79:1--79:6}, <br>
articleno = {79}, <br>
numpages = {6}, <br>
url = {http://doi.acm.org/10.1145/3009977.3010037}, <br>
doi = {10.1145/3009977.3010037}, <br>
acmid = {3010037}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
} <br>
</p>
</div>
<div id="FeatureViz_icvgipabs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
Visualizing 2-D/3-D embeddings of image features can help gain an intuitive understanding of the image category landscape. However, popular methods for visualizing such embeddings (e.g. color-coding by category) are impractical when the number of categories is large. To address this and other shortcomings, we propose novel quantitative measures defined on image feature embeddings. Each measure produces a ranked ordering of the categories and provides an intuitive vantage point from which to view the entire set of categories. As an experimental testbed, we use deep features obtained from category-epitomes, a recently introduced minimalist visual representation, across 160 object categories. We embed the features in a visualization-friendly yet similarity-preserving 2-D manifold and analyze the inter/intra-category distributions of these embeddings using the proposed measures. Our analysis demonstrates that the category ordering methods enable new insights for the domain of large-category object representations. Moreover, our ordering measure approach is general in nature and can be applied to any feature-based representation of categories.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/new_burst.png"></div>
</li>
</ul>
<p style="margin:-8px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/epitomes-to-wordle.jpg"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Analyzing Structural Characteristics of Object Category Representations From Their Semantic-part Distributions</h4>
<p style="margin:-9.0px 0px 0px 0px;"><strong>Ravi Kiran Sarvadevabhatla</strong>, R. Venkatesh Babu</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>ACM Multimedia Conference (<strong>ACMMM</strong>)</em>, Amsterdam, The Netherlands 2016</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;"> To analyze intra/inter-category variations of object apperance, we present an approach which represents the relative frequency of object part presence as a category-level word cloud. In this paper, we explore the word cloud style visualizations to characterize <i>category-epitomes</i>, a novel visual representation for objects designed by us in a previous work. </p>
<a href="http://dl.acm.org/ft_gateway.cfm?id=2967190&ftid=1796462&dwn=1&CFID=874283070&CFTOKEN=86936959" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#SketchPartsMM16abs" href="#SketchPartsMM16abs-list">Abstract</a>
<a href="http://val.serc.iisc.ernet.in/eotd/epipartviz/" target="_blank" class="buttonPP">Project page</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#SketchPartsMM16" href="#SketchPartsMM16-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="SketchPartsMM16-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Sarvadevabhatla:2016:ASC:2964284.2967190,<br>
author = {Sarvadevabhatla, Ravi Kiran and R, Venkatesh Babu}, <br>
title = {Analyzing Structural Characteristics of Object Category Representations From Their Semantic-part Distributions}, <br>
booktitle = {Proceedings of the 2016 ACM on Multimedia Conference}, <br>
series = {MM '16}, <br>
year = {2016}, <br>
isbn = {978-1-4503-3603-1}, <br>
location = {Amsterdam, The Netherlands}, <br>
pages = {97--101}, <br>
numpages = {5}, <br>
url = {http://doi.acm.org/10.1145/2964284.2967190}, <br>
doi = {10.1145/2964284.2967190}, <br>
acmid = {2967190}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
keywords = {freehand sketch, object category representation, semantic part, visualization}, <br>
}
</p>
</div>
<div id="SketchPartsMM16abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
Studies from neuroscience show that part-mapping computations are employed by human visual system in the process of object recognition. In this paper, we present an approach for analyzing semantic-part characteristics of object category representations. For our experiments, we use category-epitome, a recently proposed sketch-based spatial representation for objects. To enable part-importance analysis, we first obtain semantic-part annotations of hand-drawn sketches originally used to construct the epitomes. We then examine the extent to which the semantic-parts are present in the epitomes of a category and visualize the relative importance of parts as a word cloud. Finally, we show how such word cloud visualizations provide an intuitive understanding of category-level structural trends that exist in the category-epitome object representations. Our method is general in applicability and can also be used to analyze part-based visual object representations for other depiction methods such as photographic images.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
</li>
</ul>
<a data-toggle="collapse" data-parent="#moviebook" href="research.html#moviebook-info">close window</a>
</div>
</div>
<div id="scene-info" class="panel-collapse collapse out" style="padding:0% 3% 1% 3%; border-radius:6px; border-style: solid; border-color: #666666;">
<p style="margin:10px 0px 0px 0px;"></p>
<font style="font-size:1.34em; font-weight:500;">Affective Computing</font>
<div class="media">
<p align="justify">
Affective computing deals with, among many other things, design of models which can process human affect (emotions, interactions). In developing these models, the various human modalities (facial expression, vocal signals) tend to be treated individually or fused without considering their how the modalities affect one another. To address this, I have developed a framework for facial expression recognition which takes other modalities (head tilt, whether the person is speaking) into account. In another work, I have developed a computational model for humanoid robots which aims to mimick the way humans attend to multiple people who happen to be interacting with them.
</p>
<p style="margin:30px 0px 0px 0px;"></p>
<h3> Relevant Publications</h3>
<br>
<p style="margin:-12px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/facialexprec.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Adaptive Facial Expression Recognition Using Inter-modal Top-down context </h4>
<p style="margin:-9.0px 0px 0px 0px;"><strong>Ravi Kiran Sarvadevabhatla</strong>, Victor Ng-Thow-Hing, Mitchel Benovoy, Sam Musallam</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>International Conference on Multimodal Interaction(<strong>ICMI</strong>) </em>, Alicante, Spain 2011</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;"> This paper describes an approach for facial expression recognition which takes the effect of other concurrently active modalities (e.g. talking while emoting the expression) into account. </p>
<a href="http://dl.acm.org/ft_gateway.cfm?id=2070488&ftid=1058231&dwn=1&CFID=875199498&CFTOKEN=10876262" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#ExpRec11abs" href="#ExpRec11abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#ExpRec11" href="#ExpRec11-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="ExpRec11-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Sarvadevabhatla:2011:AFE:2070481.2070488, <br>
author = {Sarvadevabhatla, Ravi Kiran and Benovoy, Mitchel and Musallam, Sam and Ng-Thow-Hing, Victor}, <br>
title = {Adaptive Facial Expression Recognition Using Inter-modal Top-down Context}, <br>
booktitle = {Proceedings of the 13th International Conference on Multimodal Interfaces}, <br>
series = {ICMI '11}, <br>
year = {2011}, <br>
isbn = {978-1-4503-0641-6}, <br>
location = {Alicante, Spain}, <br>
pages = {27--34}, <br>
numpages = {8}, <br>
url = {http://doi.acm.org/10.1145/2070481.2070488}, <br>
doi = {10.1145/2070481.2070488}, <br>
acmid = {2070488}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
keywords = {context, facial expression recognition, human-computer interaction, mask, multi-modal}, <br>
}
</p>
</div>
<div id="ExpRec11abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
The role of context in recognizing a person's affect is being increasingly studied. In particular, context arising from the presence of multi-modal information such as faces, speech and head pose has been used in recent studies to recognize facial expressions. In most approaches, the modalities are independently considered and the effect of one modality on the other, which we call inter-modal influence (e.g. speech or head pose modifying the facial appearance) is not modeled. In this paper, we describe a system that utilizes context from the presence of such inter-modal influences to recognize facial expressions. To do so, we use 2-D contextual masks which are activated within the facial expression recognition pipeline depending on the prevailing context. We also describe a framework called the Context Engine. The Context Engine offers a scalable mechanism for extending the current system to address additional modes of context that may arise during human-machine interactions. Results on standard data sets demonstrate the utility of modeling inter-modal contextual effects in recognizing facial expressions.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/Oral-session-icon.png"></div>
</li>
</ul>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/panattn.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Panoramic Attention for Humanoid Robots</h4>
<p style="margin:-9.0px 0px 0px 0px;"> <strong>Ravi Kiran Sarvadevabhatla</strong>, Victor Ng-Thow-Hing</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>IEEE-RAS International Conference on Humanoid Robots (<strong>Humanoids</strong>)</em> Paris, France 2009</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<a href="http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5379578" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#PanAttn09abs" href="#PanAttn09abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#PanAttn09" href="#PanAttn09-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="PanAttn09-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{sarvadevabhatla2009panoramic,<br>
title={Panoramic attention for humanoid robots},<br>
author={Sarvadevabhatla, Ravi Kiran and Ng-Thow-Hing, Victor}, <br>
booktitle={2009 9th IEEE-RAS International Conference on Humanoid Robots}, <br>
pages={215--222}, <br>
year={2009}, <br>
organization={IEEE} <br>
}
</p>
</div>
<div id="PanAttn09abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
In this paper, we present a novel three-layer model of panoramic attention for our humanoid robot. In contrast to similar architectures employing coarse discretizations of the panoramic field, saliencies are maintained only for cognitively prominent entities(e.g. faces). In the absence of attention triggers, an idle-policy makes the humanoid span the visual field of panorama imparting a human-like idle gaze while simultaneously registering attention-worthy entities. We also describe a model of cognitive panoramic habituation which maintains entity-specific persistence models, thus imparting lifetimes to entities registered across the panorama. This mechanism enables the memories of entities in the panorama to fade away, creating a human-like attentional effect. We describe scenarios demonstrating the aforementioned aspects. In addition, we present experimental results which demonstrate how the cognitive filtering aspect of our model reduces processing time and false-positive rates for standard entity related modules such as face-detection and recognition.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
</li>
</ul>
<a data-toggle="collapse" data-parent="#scene" href="research.html#scene-info">close window</a>
</div>
</div>
<div id="imagetext-info" class="panel-collapse collapse out" style="padding:0% 3% 1% 3%; border-radius:6px; border-style: solid; border-color: #666666;">
<p style="margin:10px 0px 0px 0px;"></p>
<font style="font-size:1.34em; font-weight:500;">Human-Robot Interaction</font>
<div class="media">
<p align="justify">
My research in the area of Human-Robot Interaction has involved collection and rich analysis of the large-scale multi-modal data produced during interactions between humans and robots. In particular, I have designed systems to ease analysis of data from interaction sessions which last many hours. In addition, I have contributed crucial audio and video analytics components in a large-scale distributed system designed to enable Honda's humanoid robot ASIMO to play the popular card game - 'Memory Game'.
</p>
<p style="margin:30px 0px 0px 0px;"></p>
<h3> Relevant Publications</h3>
<br>
<p style="margin:-12px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/captainmayi.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%"> Captain May I -- Proxemics studies examining factors that influence distance between humanoid robots, children and adults during Human-Robot Interaction </h4>
<p style="margin:-9.0px 0px 0px 0px;">Sandra Okita, Victor Ng-Thow-Hing, <strong>Ravi Kiran Sarvadevabhatla</strong></p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>ACM/IEEE International Conference on Human-Robot Interaction (<strong>HRI</strong>)</em>, Boston, USA 2012</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;"> This paper examines how interaction distance between humans and robots vary due to factors such as age, initiator, gesture style, movement announcement. </p>
<a href="http://dl.acm.org/ft_gateway.cfm?id=2157756&ftid=1161758&dwn=1&CFID=875199498&CFTOKEN=10876262" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#Captain12abs" href="#Captain12abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#Captain12" href="#Captain12-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="Captain12-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Okita:2012:CMI:2157689.2157756, <br>
author = {Okita, Sandra Y. and Ng-Thow-Hing, Victor and Sarvadevabhatla, Ravi Kiran}, <br>
title = {Captain May I?: Proxemics Study Examining Factors That Influence Distance Between Humanoid Robots, Children, and Adults, During Human-robot Interaction}, <br>
booktitle = {Proceedings of the Seventh Annual ACM/IEEE International Conference on Human-Robot Interaction}, <br>
series = {HRI '12}, <br>
year = {2012}, <br>
isbn = {978-1-4503-1063-5}, <br>
location = {Boston, Massachusetts, USA}, <br>
pages = {203--204}, <br>
numpages = {2}, <br>
url = {http://doi.acm.org/10.1145/2157689.2157756}, <br>
doi = {10.1145/2157689.2157756}, <br>
acmid = {2157756}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
keywords = {human robot interaction, proxemics study, young children}, <br>
}
</p>
</div>
<div id="Captain12abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
This proxemics study examines whether the physical distance between robots and humans differ based on the following factors: 1) age: children vs. adults, 2) who initiates the approach: humans approaching the robot vs. robot approaching humans, 3) prompting: verbal invitation vs. non-verbal gesture (e.g., beckoning), and 4) informing: announcement vs. permission vs. nothing. Results showed that both verbal and non-verbal prompting had significant influence on physical distance. Physiological data is also used to detect the appropriate timing of approach for a more natural and comfortable interaction.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
</li>
</ul>
<p style="margin:-12.5px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/facialexprec.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Adaptive Facial Expression Recognition Using Inter-modal Top-down context </h4>
<p style="margin:-9.0px 0px 0px 0px;"><strong>Ravi Kiran Sarvadevabhatla</strong>, Victor Ng-Thow-Hing, Mitchel Benovoy, Sam Musallam</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>International Conference on Multimodal Interaction(<strong>ICMI</strong>) </em>, Alicante, Spain 2011</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px; color:#E62E00;"> This paper describes an approach for facial expression recognition which takes the effect of other concurrently active modalities (e.g. talking while emoting the expression) into account. </p>
<a href="http://dl.acm.org/ft_gateway.cfm?id=2070488&ftid=1058231&dwn=1&CFID=875199498&CFTOKEN=10876262" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#ExpRec11abs" href="#ExpRec11abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#ExpRec11" href="#ExpRec11-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="ExpRec11-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{Sarvadevabhatla:2011:AFE:2070481.2070488, <br>
author = {Sarvadevabhatla, Ravi Kiran and Benovoy, Mitchel and Musallam, Sam and Ng-Thow-Hing, Victor}, <br>
title = {Adaptive Facial Expression Recognition Using Inter-modal Top-down Context}, <br>
booktitle = {Proceedings of the 13th International Conference on Multimodal Interfaces}, <br>
series = {ICMI '11}, <br>
year = {2011}, <br>
isbn = {978-1-4503-0641-6}, <br>
location = {Alicante, Spain}, <br>
pages = {27--34}, <br>
numpages = {8}, <br>
url = {http://doi.acm.org/10.1145/2070481.2070488}, <br>
doi = {10.1145/2070481.2070488}, <br>
acmid = {2070488}, <br>
publisher = {ACM}, <br>
address = {New York, NY, USA}, <br>
keywords = {context, facial expression recognition, human-computer interaction, mask, multi-modal}, <br>
}
</p>
</div>
<div id="ExpRec11abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
The role of context in recognizing a person's affect is being increasingly studied. In particular, context arising from the presence of multi-modal information such as faces, speech and head pose has been used in recent studies to recognize facial expressions. In most approaches, the modalities are independently considered and the effect of one modality on the other, which we call inter-modal influence (e.g. speech or head pose modifying the facial appearance) is not modeled. In this paper, we describe a system that utilizes context from the presence of such inter-modal influences to recognize facial expressions. To do so, we use 2-D contextual masks which are activated within the facial expression recognition pipeline depending on the prevailing context. We also describe a framework called the Context Engine. The Context Engine offers a scalable mechanism for extending the current system to address additional modes of context that may arise during human-machine interactions. Results on standard data sets demonstrate the utility of modeling inter-modal contextual effects in recognizing facial expressions.
</p>
<p style="margin:0px 0px 0px 0px;"></p>
</div>
</div>
<div class="pull-right text-center" style="padding:4px 0.5% 0 0; width:5.0%;" href=""><img width="100%" src="./ravikiran_files/Oral-session-icon.png"></div>
</li>
</ul>
<p style="margin:-13.5px 0px 0px 0px;"></p>
<p style="margin:-12.5px 0px 0px 0px;"></p>
<ul class="list-group">
<li class="list-group-item" style="padding:0 0 0.2% 1.1%;">
<div class="media">
<div class="pull-left text-center" style="padding:15px 1.5% 0 0; width:14.1%;" href=""><img width="100%" src="./ravikiran_files/extduration.png"></div>
<div class="pull-left text-left" style="padding:0% 0% 0% 0.1%; width:76.4%;">
<p style="margin:-3px 0px 0px 0px;"></p>
<h4 style="font-size:14.1px; line-height:120%">Extended duration human-robot interaction: tools and analysis</h4>
<p style="margin:-9.0px 0px 0px 0px;"> <strong>Ravi Kiran Sarvadevabhatla</strong>, Victor Ng-Thow-Hing, Sandra Okita</p>
<p style="font-size:13.4px"></p>
<p style="margin:-10.0px 0px 0px 0px;"></p>
<p style="font-size:13.4px">In <em>IEEE International Symposium on Robot and Human Interactive Communication (<strong>RO-MAN</strong>)</em>, Viareggio, Italy 2010</p>
<p style="margin:-9.0px 0px 0px 0px;"></p>
<a href="http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5598676" class="buttonTT">Paper</a>
<a class="buttonAA" data-toggle="collapse" data-parent="#Duration10abs" href="#Duration10abs-list">Abstract</a>
<a class="buttonSS" data-toggle="collapse" data-parent="#Duration10" href="#Duration10-list">Bibtex</a><br>
<p style="margin:1px 0px 0px 0px;"></p>
<div id="Duration10-list" class="panel-collapse collapse out" style="background-color:#FFE0C2; padding:0% 3% 1% 3%; border-radius:10px;">
<p style="font-size:15px;">
@inproceedings{sarvadevabhatla2010extended, <br>
title={Extended duration human-robot interaction: tools and analysis}, <br>
author={Sarvadevabhatla, Ravi Kiran and Ng-Thow-Hing, Victor and Okita, Sandra},<br>
booktitle={19th International Symposium in Robot and Human Interactive Communication}, <br>
pages={7--14}, <br>
year={2010}, <br>
organization={IEEE} <br>
}
</p>
</div>
<div id="Duration10abs-list" class="panel-collapse collapse out" style="background-color:#ADEBFF; padding:9px 2.5% 3px 2.5%; border-radius:10px;">
<p style="margin:0px 0px 0px 0px;"></p>
<p style="font-size:14.1px;">
Extended human-robot interactions possess
unique aspects which are not exhibited in short-term
interactions spanning a few minutes or extremely long-term
spanning days. In order to comprehensively monitor such
interactions, we need special recording mechanisms which
ensure the interaction is captured at multiple spatio-temporal
scales, viewpoints and modalities(audio, video, physio). To
minimize cognitive burden, we need tools which can automate
the process of annotating and analyzing the resulting data. In
addition, we also require these tools to be able to provide a
unified, multi-scale view of the data and help discover patterns
in the interaction process. In this paper, we describe recording
and analysis tools which are helping us analyze extended
human-robot interactions with children as subjects. We also
provide some experimental results which highlight the utility
of such tools.
</p>
<p style="margin:0px 0px 0px 0px;"></p>