0.25
0.5
0.75
1.25
1.5
1.75
2
SSD: Single Shot MultiBox Detector
Published on Oct 24, 20165573 Views
We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over differe
Related categories
Chapter list
SSD: Single Shot MultiBox Detector00:00
Demo video - Faster than real time on Titan X00:08
Speed and accuracy - 100:36
Speed and accuracy - 200:58
Speed and accuracy - 301:05
Speed and accuracy - 401:18
Two-Stage versus Single Shot01:32
Bounding Box Prediction - 102:20
Bounding Box Prediction - 202:23
Bounding Box Prediction - 302:36
Bounding Box Prediction - 402:57
Bounding Box Prediction - 503:04
Bounding Box Prediction - 603:07
Bounding Box Prediction - 703:08
Bounding Box Prediction - 803:09
Bounding Box Prediction - 903:11
Bounding Box Prediction - 1003:13
SSD Output Layer - 103:15
SSD Output Layer - 203:33
SSD Output Layer - 303:44
SSD Training04:20
Related Work - 104:38
Related Work - 205:07
Related Work - 305:15
Related Work - 405:31
Related Work - 505:51
Related Work - 606:00
Contribution #1: Multi-Scale Feature Maps - 106:07
Contribution #1: Multi-Scale Feature Maps - 206:17
Contribution #1: Multi-Scale Feature Maps - 306:23
Multi-Scale Feature Maps - 106:33
Multi-Scale Feature Maps - 206:42
Multi-Scale Feature Maps Experiment - 107:11
Multi-Scale Feature Maps Experiment - 207:20
Multi-Scale Feature Maps Experiment - 307:28
Multi-Scale Feature Maps Experiment - 407:50
Multi-Scale Feature Maps Experiment - 507:52
Multi-Scale Feature Maps Experiment - 607:55
Contribution #2: Splitting the Region Space - 107:59
Contribution #2: Splitting the Region Space - 208:21
Contribution #2: Splitting the Region Space - 308:34
Why So Many Default Boxes? - 108:47
Why So Many Default Boxes? - 209:01
Why So Many Default Boxes? - 309:06
Why So Many Default Boxes? - 409:10
Why So Many Default Boxes? - 509:14
Why So Many Default Boxes? - 609:22
Why So Many Default Boxes? - 709:25
Handling Many Default Boxes - 109:46
Handling Many Default Boxes - 210:04
Handling Many Default Boxes - 310:09
Handling Many Default Boxes - 410:20
Handling Many Default Boxes - 5410:35
Handling Many Default Boxes - 610:37
Handling Many Default Boxes - 710:45
SSD Architecture10:54
Contribution #3: The Devil is in the Details11:15
Data Augmentation - 111:33
Data Augmentation - 211:50
Data Augmentation - 312:10
Results on VOC2007 test - 112:18
Results on VOC2007 test - 212:20
Results on VOC2007 test - 312:21
Results on VOC2007 test - 612:24
Results on VOC2007 test - 712:36
Results on VOC2007 test - 812:37
Results on More Datasets - 112:38
Results on More Datasets - 212:39
Results on More Datasets - 312:49
Results on More Datasets - 412:51
Results on More Datasets - 512:56
COCO Bounding Box precision - 113:00
COCO Bounding Box precision - 213:01
Future Work - 113:17
Future Work - 213:20
Future Work - 313:36
Future Work - 413:39
Check out the code/models13:52
Thank you!13:57