Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation •...
Transcript of Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation •...
![Page 2: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/2.jpg)
Visual Recognition
A fundamental task in computer vision
• Classification
• Object Detection
• Semantic Segmentation
• Instance Segmentation
• Key point Detection
• VQA
…
![Page 3: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/3.jpg)
Category-level Recognition
Category-level Recognition Instance-level Recognition
![Page 4: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/4.jpg)
Representation
• Bounding-box• Face Detection, Human Detection, Vehicle Detection, Text Detection,
general Object Detection• Point
• Semantic segmentation (will be discussed in next week)• Keypoint
• Face landmark• Human Keypoint
![Page 5: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/5.jpg)
Outline
• Detection• Human Keypoint• Conclusion
![Page 6: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/6.jpg)
Outline
• Detection• Human Keypoint• Conclusion
![Page 7: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/7.jpg)
Detection - Evaluation CriteriaAverage Precision (AP) and mAP
Figures are from wikipedia
![Page 8: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/8.jpg)
Detection - Evaluation CriteriammAP
Figures are from http://cocodataset.org
![Page 9: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/9.jpg)
How to perform a detection?
• Sliding window: enumerate all the windows (up to millions of windows)• VJ detector: cascade chain
• Fully Convolutional network• shared computation
Robust Real-time Object Detection; Viola, Jones; IJCV 2001
http://www.vision.caltech.edu/html-files/EE148-2005-Spring/pprs/viola04ijcv.pdf
![Page 10: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/10.jpg)
General Detection Before Deep Learning
• Feature + classifier• Feature
• Haar Feature• HOG (Histogram of Gradient)• LBP (Local Binary Pattern)• ACF (Aggregated Channel Feature)• …
• Classifier• SVM• Bootsing• Random Forest
![Page 11: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/11.jpg)
Traditional Hand-crafted Feature: HoG
![Page 12: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/12.jpg)
Traditional Hand-crafted Feature: HoG
![Page 13: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/13.jpg)
General Detection Before Deep Learning
Traditional Methods• Pros
• Efficient to compute (e.g., HAAR, ACF) on CPU• Easy to debug, analyze the bad cases• reasonable performance on limited training data
• Cons• Limited performance on large dataset• Hard to be accelerated by GPU
![Page 14: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/14.jpg)
Deep Learning for Object Detection
Based on the whether following the “proposal and refine” • One Stage
• Example: Densebox, YOLO (YOLO v2), SSD, Retina Net• Keyword: Anchor, Divide and conquer, loss sampling
• Two Stage• Example: RCNN (Fast RCNN, Faster RCNN), RFCN, FPN, MaskRCNN• Keyword: speed, performance
![Page 15: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/15.jpg)
A bit of History
ImageFeature
Extractor
classification
localization
(bbox)
One stage detector
Densebox (2015) UnitBox (2016) EAST (2017)
YOLO (2015) Anchor Free
Anchor importedYOLOv2 (2016)
SSD (2015)
RON(2017)
RetinaNet(2017)
DSSD (2017)
two stages detector
ImageFeature
Extractor
classification
localization
(bbox)
Proposal
classification
localization
(bbox)
Refine
RCNN (2014) Fast RCNN(2015)Faster RCNN (2015)
RFCN (2016)
MultiBox(2014)
RFCN++ (2017)
FPN (2017)
Mask RCNN (2017)
OverFeat(2013)
![Page 16: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/16.jpg)
One Stage Detector: Densebox
DenseBox: Unifying Landmark Localization with End to End Object Detection, Huang etc, 2015
https://arxiv.org/abs/1509.04874
![Page 17: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/17.jpg)
One Stage Detector: Densebox
• No Anchor: GT Assignment• A sub-circle in the GT is labeled as positive
• fail when two GT highly overlaps• the size of the sub-circle matters• more attention (loss) will be placed to large faces
• Loss sampling• All pos/negative positions will be used to compute the cls loss
![Page 18: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/18.jpg)
One Stage Detector: Densebox
Problems• L2 loss is not robust to scale variation (UnitBox)
• learnt features are not robust• GT assignment issue (SSD)
• Fail to handle the crowd case• relatively large localization error (Two stages detector)• more false positive (FP) (Two stages detector)
• does not obviously kill the fp
![Page 19: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/19.jpg)
One Stage Detector: Densebox -> UnitBox
UnitBox: An Advanced Object Detection Network
Yu etc, 2016
http://cn.arxiv.org/pdf/1608.01471.pdf
![Page 20: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/20.jpg)
One Stage Detector: Densebox -> UnitBox->EAST
EAST: An Efficient and Accurate Scene Text Detector, Zhou etc, CVPR 2017
https://arxiv.org/abs/1704.03155
![Page 21: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/21.jpg)
One Stage Detector: YOLO
You Only Look Once: Unified, Real-Time Object Detection, Redmon etc, CVPR 2016
https://arxiv.org/abs/1506.02640
![Page 22: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/22.jpg)
One Stage Detector: YOLO
You Only Look Once: Unified, Real-Time Object Detection, Redmon etc, CVPR 2016
https://arxiv.org/abs/1506.02640
![Page 23: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/23.jpg)
One Stage Detector: YOLO
• No Anchor • GT assignment is based on the cells (7x7)
• Loss sampling• all pos/neg predictions are evaluated (but more sparse than densebox)
You Only Look Once: Unified, Real-Time Object Detection, Redmon etc, CVPR 2016
https://arxiv.org/abs/1506.02640
![Page 24: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/24.jpg)
One Stage Detector: YOLO
Discussion• fc reshape (4096-> 7x7x30)
• more context• but not fully convolutional
• One cell can output up to two boxes in one category • fail to work on the crowd case
• Fast speed• small imagenet base model• small input size (448x448)
You Only Look Once: Unified, Real-Time Object Detection, Redmon etc, CVPR 2016
https://arxiv.org/abs/1506.02640
![Page 25: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/25.jpg)
One Stage Detector: YOLO
Experiments on general detection
You Only Look Once: Unified, Real-Time Object Detection, Redmon etc, CVPR 2016
https://arxiv.org/abs/1506.02640
Method VOC 2007 test VOC 2012 test COCO time
YOLO 57.9/NA 52.7/63.4 NA fps: 45/155
![Page 26: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/26.jpg)
One Stage Detector: YOLO -> YOLOv2
YOLO9000: Better, Faster, Stronger Redmon etc, CVPR 2016
https://arxiv.org/abs/1612.08242
![Page 27: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/27.jpg)
One Stage Detector: YOLO -> YOLOv2
Experiments:
YOLO9000: Better, Faster, Stronger Redmon etc, CVPR 2016
https://arxiv.org/abs/1612.08242
Method VOC 2007 test VOC 2012 test COCO time
YOLO 52.7/63.4 57.9/NA NA fps: 45/155
YOLOv2 78.6 73.4 21.6 fps: 40
![Page 28: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/28.jpg)
One Stage Detector: YOLO -> YOLOv2
Video demo: https://pjreddie.com/darknet/yolo/
YOLO9000: Better, Faster, Stronger Redmon etc, CVPR 2016
https://arxiv.org/abs/1612.08242
![Page 29: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/29.jpg)
One Stage Detector: SSD
SSD: Single Shot MultiBox Detector, Liu etc
https://arxiv.org/pdf/1512.02325.pdf
![Page 30: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/30.jpg)
One Stage Detector: SSD
SSD: Single Shot MultiBox Detector, Liu etc, ECCV 2016
https://arxiv.org/pdf/1512.02325.pdf
![Page 31: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/31.jpg)
One Stage Detector: SSD• Anchor
• GT-anchor assignment• GT is predicted by one best matched (IOU) anchor or matched with an
anchor with IOU > 0.5• better recall
• dense or sparse anchor?• Divide and Conquer
• Different layers handle the objects with different scales• Assume small objects can be predicted in earlier layers (not very strong
semantics)• Loss sampling
• OHEM: negative positions are sampled (not balanced pos/neg ratio)• negative:pos is at most 3:1
SSD: Single Shot MultiBox Detector, Liu etc, ECCV 2016
https://arxiv.org/pdf/1512.02325.pdf
![Page 32: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/32.jpg)
One Stage Detector: SSD
Discussion:• Assume small objects can be predicted in earlier layers (not very
strong semantics) (DSSD, RON, RetinaNet)• strong data augmentation• VGG model (Replace by resnet in DSSD)
• cannot be easily adapted to other models• a lot of hacks
• A long tail (Large computation)
SSD: Single Shot MultiBox Detector, Liu etc, ECCV 2016
https://arxiv.org/pdf/1512.02325.pdf
![Page 33: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/33.jpg)
One Stage Detector: SSD
Experiments
SSD: Single Shot MultiBox Detector, Liu etc, ECCV 2016
https://arxiv.org/pdf/1512.02325.pdf
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
![Page 34: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/34.jpg)
One Stage Detector: SSD -> DSSD
DSSD : Deconvolutional Single Shot Detector, Fu etc 2017,
https://arxiv.org/abs/1701.06659
![Page 35: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/35.jpg)
One Stage Detector: DSSD
Experiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
DSSD : Deconvolutional Single Shot Detector, Fu etc 2017,
https://arxiv.org/abs/1701.06659
![Page 36: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/36.jpg)
One Stage Detector: SSD -> RON
RON: Reverse Connection with Objectness Prior Networks for Object Detection, Kong etc, CVPR 2017
https://arxiv.org/pdf/1707.01691.pdf
![Page 37: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/37.jpg)
One Stage Detector: RON
• Anchor
• Divide and conquer
• Reverse Connect (similar to FPN)
• Loss Sampling
• Objectness prior
• pos/neg unbalanced issue
• split to 1) binary cls 2) multi-class cls
RON: Reverse Connection with Objectness Prior Networks for Object Detection, Kong etc, CVPR 2017
https://arxiv.org/pdf/1707.01691.pdf
![Page 38: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/38.jpg)
One Stage Detector: RON
Experiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RON: Reverse Connection with Objectness Prior Networks for Object Detection, Kong etc, CVPR 2017
https://arxiv.org/pdf/1707.01691.pdf
![Page 39: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/39.jpg)
One Stage Detector: SSD -> RetinaNet
Focal Loss for Dense Object Detection, Lin etc, ICCV 2017
https://arxiv.org/pdf/1708.02002.pdf
![Page 40: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/40.jpg)
One Stage Detector: SSD -> RetinaNet
Focal Loss for Dense Object Detection, Lin etc, ICCV 2017
https://arxiv.org/pdf/1708.02002.pdf
![Page 41: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/41.jpg)
One Stage Detector: RetinaNet
• Anchor
• Divide and Conquer
• FPN
• Loss Sampling
• Focal loss
• pos/neg unbalanced issue
• new setting (e.g., more anchor)
Focal Loss for Dense Object Detection, Lin etc, ICCV 2017
https://arxiv.org/pdf/1708.02002.pdf
![Page 42: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/42.jpg)
One Stage Detector: RetinaNet
Experiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
Focal Loss for Dense Object Detection, Lin etc, ICCV 2017
https://arxiv.org/pdf/1708.02002.pdf
![Page 43: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/43.jpg)
One Stage Detector: Summary
• Anchor• No anchor: YOLO, densebox/unitbox/east • Anchor: YOLOv2, SSD, DSSD, RON, RetinaNet
• Divide and conquer• SSD, DSSD, RON, RetinaNet
• loss sample• all sample: densebox• OHEM: SSD• focal loss: RetinaNet
![Page 44: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/44.jpg)
One Stage Detector: DiscussionAnchor (YOLO v2, SSD, RetinaNet) or Without Anchor (Densebox, YOLO)• Model Complexity
• Difference on the extremely small model (< 30M flops on 224x224 input)• Sampling • Application
• No Anchor: Face• With Anchor: Human, General Detection
• Problem for one stage detector• Unbalanced pos/neg data• Pool localization precision
![Page 45: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/45.jpg)
Two Stages Detector: RCNN
Rich feature hierarchies for accurate object detection and semantic segmentation, Girshirk etc, CVPR 2014
https://arxiv.org/pdf/1311.2524.pdf
![Page 46: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/46.jpg)
Two Stages Detector: RCNN
Discussion• Extremely slow speed
• selective search proposal (CPU)/warp• not end-to-end optimized• Good for small objects
Rich feature hierarchies for accurate object detection and semantic segmentation, Girshirk etc, CVPR 2014
https://arxiv.org/pdf/1311.2524.pdf
![Page 47: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/47.jpg)
Two Stages Detector: RCNN
Experiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
RCNN 66 NA NA 47s
Rich feature hierarchies for accurate object detection and semantic segmentation, Girshirk etc, CVPR 2014
https://arxiv.org/pdf/1311.2524.pdf
![Page 48: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/48.jpg)
Two Stages Detector: RCNN -> Fast RCNN
Fast R-CNN, Girshick etc, ICCV 2015
https://arxiv.org/pdf/1504.08083.pdf
![Page 49: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/49.jpg)
Two Stages Detector: Fast RCNN
Discussion• slow speed
• selective search proposal (CPU)• not end-to-end optimized• ROI pooling
• alignment issue• sampling• aspect ratio changes
Fast R-CNN, Girshick etc, ICCV 2015
https://arxiv.org/pdf/1504.08083.pdf
![Page 50: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/50.jpg)
Two Stages Detector: Fast RCNN
Experiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
RCNN 66 NA NA 47s
Fast RCNN 77.0 82.3 (wth coco data) NA 0.5s
Fast R-CNN, Girshick etc, ICCV 2015
https://arxiv.org/pdf/1504.08083.pdf
![Page 51: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/51.jpg)
Two Stages Detector: RCNN -> Fast RCNN -> FasterRCNN
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, Ren etc, CVPR 2016
https://arxiv.org/pdf/1506.01497.pdf
![Page 52: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/52.jpg)
Two Stages Detector: Faster RCNN
Discussion• speed
• selective search proposal (CPU) -> RPN• alternative optimization/end-to-end optimization
• Recall issue due to two stages detector
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, Ren etc, CVPR 2016
https://arxiv.org/pdf/1506.01497.pdf
![Page 53: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/53.jpg)
Two Stages Detector: Faster RCNNExperiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
RCNN 66 NA NA 47s
Fast RCNN 77.0 82.3 (wth coco data) NA 0.5s
Faster RCNN 73.2 70.4 NA 5
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, Ren etc, CVPR 2016
https://arxiv.org/pdf/1506.01497.pdf
![Page 54: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/54.jpg)
Two Stages Detector: RCNN -> Fast RCNN -> FasterRCNN -> RFCN
R-FCN: Object Detection via Region-based Fully Convolutional Networks, Dai etc, NIPS 2016,
https://arxiv.org/pdf/1605.06409.pdf
![Page 55: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/55.jpg)
Two Stages Detector: RFCN
Discussion
• Share convolution
• fasterRCNN: shared Res1-4 (RPN), not shared Res5 (RCNN)
• RFCN: shared Res1-5 (both RPN and RCNN)
• PSPooling
• a large number of channels:(7x7xC)xWxH
• Problems in ROIPooling also exist
• Fully connected vs Convolution
• fc: global context
• conv: can be shared but the context is relative small
• trade-off: large kernelR-FCN: Object Detection via Region-based Fully Convolutional Networks, Dai etc, NIPS 2016,
https://arxiv.org/pdf/1605.06409.pdf
![Page 56: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/56.jpg)
Two Stages Detector: RFCNExperiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
RCNN 66 NA NA 47s
Fast RCNN 77.0 82.3 (wth coco data) NA 0.5s
Faster RCNN 73.2 70.4 NA 200ms
RFCN 79.5 77.6 29.9 170ms
R-FCN: Object Detection via Region-based Fully Convolutional Networks, Dai etc, NIPS 2016,
https://arxiv.org/pdf/1605.06409.pdf
![Page 57: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/57.jpg)
Two Stages Detector: RFCN -> Deformable Convolutional Networks
Deformable Convolutional Networks, Dai etc, ICCV 2017
https://arxiv.org/abs/1703.06211
![Page 58: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/58.jpg)
Two Stages Detector: RFCN -> Deformable Convolutional Networks
Deformable Convolutional Networks, Dai etc, ICCV 2017
https://arxiv.org/abs/1703.06211
![Page 59: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/59.jpg)
Two Stages Detector: RFCN -> Deformable Convolutional Networks
Discussion• Deformable pool is similar to ROIAlign (in Mask RCNN)• Deformable conv
• flexible to learn the non-rigid objects
Deformable Convolutional Networks, Dai etc, ICCV 2017
https://arxiv.org/abs/1703.06211
![Page 60: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/60.jpg)
Two Stages Detector: RCNN -> Fast RCNN -> FasterRCNN -> FPN
Feature Pyramid Networks for Object Detection, Lin etc, CVPR 2017
https://arxiv.org/pdf/1612.03144.pdf
![Page 61: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/61.jpg)
Two Stages Detector: FPN
Discussion
• FasterRCNN reproduced (setting)
• Deeply supervised (better feature)
Feature Pyramid Networks for Object Detection, Lin etc, CVPR 2017
https://arxiv.org/pdf/1612.03144.pdf
![Page 62: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/62.jpg)
Two Stages Detector: FPNExperiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
RCNN 66 NA NA 47s
Fast RCNN 77.0 82.3 (wth coco data) NA 0.5s
Faster RCNN 73.2 70.4 NA 200ms
RFCN 79.5 77.6 29.9 170ms
FPN NA NA 36.2 6
Feature Pyramid Networks for Object Detection, Lin etc, CVPR 2017
https://arxiv.org/pdf/1612.03144.pdf
![Page 63: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/63.jpg)
Two Stages Detector: RCNN -> Fast RCNN -> FasterRCNN -> FPN -> MaskRCNN
Mask R-CNN, He etc, ICCV 2017
https://arxiv.org/pdf/1703.06870.pdf
![Page 64: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/64.jpg)
Two Stages Detector: RCNN -> Fast RCNN -> FasterRCNN -> FPN -> MaskRCNN
Mask R-CNN, He etc, ICCV 2017
https://arxiv.org/pdf/1703.06870.pdf
![Page 65: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/65.jpg)
Two Stages Detector: Mask RCNN
Discussion
• Alignment issue in ROIPooling -> ROIAlign
• Multi-task learning: detection & mask
Mask R-CNN, He etc, ICCV 2017
https://arxiv.org/pdf/1703.06870.pdf
![Page 66: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/66.jpg)
Two Stages Detector: Mask RCNNExperiments
Method VOC 2007 test VOC 2012 test COCO time (fps)
YOLO 52.7/63.4 57.9/NA NA 45/155
YOLOv2 78.6 73.4 21.6 40
SSD 77.2/79.8 75.8/78.5 25.1/28.8 46/19
DSSD 81.5 80.0 33.2 5.5
RON 81.3 80.7 27.4 15
RetinaNet NA N 39.1 5
RCNN 66 NA NA 47s
Fast RCNN 77.0 82.3 (wth coco data) NA 0.5s
Faster RCNN 73.2 70.4 NA 200ms
RFCN 79.5 77.6 29.9 170ms
FPN NA NA 36.2 6
Mask RCNN NA NA 38.2 2.5
Mask R-CNN, He etc, ICCV 2017
https://arxiv.org/pdf/1703.06870.pdf
![Page 67: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/67.jpg)
Two Stages Detector: Summary
• Speed• RCNN -> Fast RCNN -> Faster RCNN -> RFCN
• performance• Divide and conquer
• FPN• Deformable Pool/ROIAlign• Deformable Conv• Multi-task learning
![Page 68: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/68.jpg)
Two Stages Detector: Discussion
FasterRCNN vs RFCNOne stage vs two Stage
![Page 69: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/69.jpg)
MegDetection
Introduction & Demo Video
![Page 70: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/70.jpg)
Open Problem in Detection
• FP• NMS (detection in crowd)• GT assignment issue• Detection in video
• detect & track in a network
![Page 71: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/71.jpg)
Outline
• Detection
• Human Keypoint• Conclusion
![Page 72: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/72.jpg)
Human Keypoint Task
• Single Person Skeleton• Cropped RGB image -> 2d key points / 3d key points• Keyword: inter-middle loss, large receptive field, context
• Multiple-Person Skeleton• RGB image -> human localization & human Keypoint for each person
![Page 73: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/73.jpg)
Single Person Skeleton: CPM
Convolutional Pose Machines, Wei etc, CVPR 2016
https://arxiv.org/pdf/1602.00134.pdf
![Page 74: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/74.jpg)
Single Person Skeleton: Hourglass
Stacked Hourglass Networks for Human Pose Estimation, Newell etc, ECCV 2016
https://arxiv.org/pdf/1603.06937.pdf
![Page 75: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/75.jpg)
Multiple-Person Skeleton
• Top Down• Detect -> Single person skeleton
• Bottom Up• Deep/Deeper Cut• OpenPose• Associative Embedding
![Page 76: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/76.jpg)
Multiple-Person Skeleton: OpenPose
CPM + PAF
Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields, Cao etc, CVPR 2017
https://arxiv.org/pdf/1611.08050.pdf
![Page 77: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/77.jpg)
Multiple-Person Skeleton: OpenPose
Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields, Cao etc, CVPR 2017
https://arxiv.org/pdf/1611.08050.pdf
https://github.com/CMU-Perceptual-Computing-Lab/openpose
![Page 78: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/78.jpg)
Multiple-Person Skeleton: Associative Embedding
Hourglass + AE
Associative Embedding: End-to-End Learning for Joint Detection and Grouping, Newell etc, NIPS 2017
https://arxiv.org/pdf/1611.05424.pdf
![Page 79: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/79.jpg)
Multiple-Person Skeleton: Associative Embedding
Associative Embedding: End-to-End Learning for Joint Detection and Grouping, Newell etc, NIPS 2017
https://arxiv.org/pdf/1611.05424.pdf
![Page 80: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/80.jpg)
Multiple-Person Skeleton: Discussion
• Top Down:• Depends on the detector
• Fail in the crowd case• Fail with partial observation• can detect the small-scale human
• More computation• Better localization when the input-size of single person skeleton is large
• Bottom up:• Fast computational speed• good at localizing the human with partial observation• Hard to assemble human
![Page 81: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/81.jpg)
Challenges in Skeleton
• combine top-down approaches with bottom-up approaches
• perform pose track
• handle the crowd case
![Page 82: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/82.jpg)
MegSkeleton
Introduction and demo Video
![Page 83: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/83.jpg)
Outline
• Detection
• Human Keypoint• Conclusion
![Page 84: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/84.jpg)
Conclusion
• Detection• One stage: Densebox, YOLO, SSD, RetinaNet• Two Stage: RCNN, Fast RCNN, FasterRCNN, RFCN, FPN, Mask RCNN
• Skeleton• Single Person Skeleton: CPM, Hourglass• Multi-person Skeleton
• Top Down• Bottom up: Openpose, Associative Embedding
![Page 85: Lecture 6: Modern Object Detection - zsc.github.ioObject Detection).pdf · Representation • Bounding-box • Face Detection, Human Detection, Vehicle Detection, Text Detection,](https://reader033.fdocuments.net/reader033/viewer/2022050716/5e1c28ced445df7cf4544a4c/html5/thumbnails/85.jpg)
Thanks