![Page 1: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/1.jpg)
Towards Weakly Supervised
Object Segmentation & Scene Parsing
Yunchao Wei
IFP, Beckman Institute, University of Illinois at Urbana-Champaign, IL, USA
![Page 2: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/2.jpg)
Self-Erasing Network for Integral Object AttentionQibin Hou1, Peng-Tao Jiang1, Yunchao Wei 2, Ming-Ming Cheng1
1College of Computer Science, Nankai University, Beijing, China
2IFP, Beckman Institute, University of Illinois at Urbana-Champaign, IL, USA
Image Object Localization Map
![Page 3: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/3.jpg)
…
Fully Convolutional Networks
Background
Weak Supervision: Lower degree (or cheaper, simper) annotations at training stage than the required outputs at the testing stage.
scribblespointsimage-level labels
horse
person
bounding boxes
horse person
Weak Supervision
![Page 4: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/4.jpg)
Background
FCN
LOSS
bkg
horse
person
Learn to Produce Pseudo Mask
The Popular Pipeline
![Page 5: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/5.jpg)
dog bird
Dense and integral object localization maps
Our target
Background
Small and sparse object localization maps
Class Activation Mapping [Zhou CVPR16]
Our Target & Current Issue
![Page 6: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/6.jpg)
Revisit Adversarial Erasing
Object Region Mining with Adversarial Erasing [Wei CVPR17]
![Page 7: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/7.jpg)
Revisit Adversarial Erasing
Adversarial Complementary Learning [Zhang CVPR18]
![Page 8: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/8.jpg)
Revisit Adversarial Erasing
Over Erasing: The Failure Case of Adversarial Erasing
The backgroundregions can notbe suppressed!!
![Page 9: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/9.jpg)
Our Solution: Self-Erasing Network
Motivation
Image Attention Map Ternary Mask
object priors
background priors
![Page 10: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/10.jpg)
Our Solution: Self-Erasing Network
Ternary Mask
object priors
background priors
𝛿ℎ𝛿𝑙
Attention Map
![Page 11: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/11.jpg)
Backbone
dining table
Backbone
Conv Attention Ternary Mask T
Lo
ss
Conv
GAP
SA
SB
GAP
Lo
ss...... L
oss
Conv
GAP
SC
...C-ReLU strategy I
T
T ThresholdingGAP Global Ave Pool
dining table
C-ReLU strategy II
Our Solution: Self-Erasing Network
Framework
Conv
SA
GAP
Loss...
Attention Ternary Mask T
T
Loss
Conv
GAP ... Loss
Conv
GAP
SC
...
![Page 12: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/12.jpg)
Experimental Results
Ours ACoL [Zhang CVPR18] ACoL [Zhang CVPR18]Ours
![Page 13: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/13.jpg)
Experimental Results
Pascal VOC 2012
![Page 14: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/14.jpg)
Experimental Results
![Page 15: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/15.jpg)
015
Weakly Supervised Scene Parsing with Point-based
Distance Metric LearningRui Qian1,3, Yunchao Wei 3, Honghui Shi 2,3, Jiachen Li 3, Jiaying Liu1 and Thomas Huang3
1Institute of Computer Science and Technology, Peking University, Beijing, China
2IBM T.J. Waston Research Center, 3IFP, Beckman, UIUC
Point Annotation Full Annotation
![Page 16: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/16.jpg)
016 Review on Weakly Supervised Methods
◼ Weakly supervised methods for scene parsing◼ Image-level
◼ Box supervision
◼ Scribble supervision
◼ Point supervision
PersonBikeTreeSkyRoad
![Page 17: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/17.jpg)
03 Annotation Comparison
◼ Annotation burden comparison
Method Full Scribble Point
Average
Anno.pixel/Image170K 1817.48 12.26
![Page 18: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/18.jpg)
018 Motivation
Input
PointSup
Limited
label!
Input
PointSup
Limited
label!
How to utilize limited annotation?
Cross-image semantic similarity!
train
track
tree
![Page 19: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/19.jpg)
05 Proposed method(1/5)
◼ Overview◼ Point-based distance metric learning(PDML)
◼ Point supervision(PointSup)
◼ Online extension supervision(ExtendSup)
Feature
Extract
PDML
Classification
Module
Input Feature Prediction
PointSup
ExtendSup
CrossEntropy
Loss
![Page 20: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/20.jpg)
06 Proposed method(2/5)
◼ Point supervision◼ Only calculate cross-entropy loss on annotated pixels
◼ Back propagate gradients accordingly
◼ Optimize by stochastic gradient descent
Feature
ExtractClassification
Module
Input Feature PredictionPointSup
Cro
ssE
ntro
py
![Page 21: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/21.jpg)
07 Proposed method(3/5)
◼ Online extension supervision◼ Extension method1 (region):
◼ Select pixels in 5*5 square near the annotated ones
◼ Extension method2 (score):
◼ Select pixels with score over 0.7 in the prediction
◼ Finally choose the intersection of two methods
Feature
ExtractClassification
Module
Input Feature Prediction
PointSup
Cro
ssE
ntro
py
Online
Extension
![Page 22: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/22.jpg)
08 Proposed method(4/5)
◼ Point-based distance metric learning
Shared
weights
(b)
Feature
Extract
𝐼𝑎
𝐼𝑏
Feature
Extract
Feature
Feature
platformtrain
minimizing
maximizing
(a)
Subgroup
Point-based Distance
Metric Learning
𝐼𝑎
𝐼𝑏𝐼𝑐
𝐼𝑑
train platform
![Page 23: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/23.jpg)
09 Proposed method(5/5)
◼ Loss function of PDML◼ For each image 𝐼𝑎 , define the embedding vector set as 𝐸𝑎:
◼ 𝐸𝑎 = 𝑖=1ڂ|𝑀𝑎|{𝑃𝑎𝑖}
◼ |𝑀𝑎| is the number of annotated pixel of 𝐼𝑎◼ 𝑃𝑎𝑖 is the feature vector of 𝑖th pixel
◼ We optimize in the triplet form of {𝑃𝑎𝑖 , 𝑃𝑏𝑗, 𝑃𝑏𝑘}:
◼ 𝑃𝑎𝑖 shares the same category with 𝑃𝑏𝑗◼ 𝑃𝑏𝑘 shares different category with 𝑃𝑎𝑖 , 𝑃𝑏𝑗
◼ We use the loss function of :
𝐿𝑡 𝑃𝑎𝑖 , 𝑃𝑏𝑗, 𝑃𝑏𝑘 = α𝐿𝑝 𝑃𝑎𝑖, 𝑃𝑏𝑗 + 𝛽𝐿𝑛 𝑃𝑎𝑖 , 𝑃𝑏𝑗 , 𝑃𝑏𝑘
◼ 𝐿𝑝 𝑃𝑎𝑖 , 𝑃𝑏𝑗 = ||𝑃𝑎𝑖 − 𝑃𝑏𝑗||2
◼ 𝐿𝑛 𝑃𝑎𝑖, 𝑃𝑏𝑗 , 𝑃𝑏𝑘 = max(||𝑃𝑎𝑖 − 𝑃𝑏𝑗||2 - ||𝑃𝑎𝑖 − 𝑃𝑏𝑘||2 + 𝑚, 0)
◼ α, 𝛽,𝑚 are hyper-params and are set to 0.8, 1, 20 in practice
![Page 24: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/24.jpg)
10 Datasets
◼ Scene parsing datasets◼ PASCAL-Context
◼ ADE 20K
Dataset #Training #Evaluation #Instance/Image
PASCAL-
Context4998 5105 12.26
ADE20K 20210 2000 13.96
![Page 25: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/25.jpg)
25 Experimental Results
Method Metrics
FullSup PointSup PDML Online Ext. mIoU Pixel Acc
√ 39.6 78.6%
√ 27.9 55.3%
√ √ 29.7 57.5%
√ √ √ 30.0 57.6%
◼ Quantitative evaluation on PASCAL-Context◼ The combination of three techniques is best
◼ We use only 0.007% annotated data but reached
75% of the full supervision performance!
![Page 26: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/26.jpg)
26 Experimental Results
Method Metrics
FullSup PointSup PDML Online Ext. mIoU Pixel Acc
√ 33.9 75.8%
√(SegNet) 21.0 /
√ 17.7 58.0%
√ √ 19.0 59.0%
√ √ √ 19.6 61.0%
◼ Quantitative evaluation on ADE20K◼ The combination of three techniques is best
◼ Our method approaches the result SegNet under
full supervision scheme
![Page 27: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/27.jpg)
27
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 28: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/28.jpg)
28
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 29: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/29.jpg)
29
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 30: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/30.jpg)
30
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 31: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/31.jpg)
31
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 32: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/32.jpg)
32
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 33: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/33.jpg)
33
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 34: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/34.jpg)
34
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 35: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/35.jpg)
35
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 36: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/36.jpg)
36
Image
PointSup
PDML
Final
GT
Subjective Evaluation
![Page 37: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/37.jpg)
23 Discussion
◼ Visualization the effect of PDML ◼ dis(+): L2 norm distance between same-class feature vectors
◼ dis(-): L2 norm distance between different-class feature vectors
![Page 38: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/38.jpg)
23 Discussion
◼ Ablation on the design of PDML loss function◼ 𝐿𝑡 𝑃𝑎𝑖, 𝑃𝑏𝑗 , 𝑃𝑏𝑘 = α𝐿𝑝 𝑃𝑎𝑖 , 𝑃𝑏𝑗 + 𝛽𝐿𝑛 𝑃𝑎𝑖 , 𝑃𝑏𝑗 , 𝑃𝑏𝑘
![Page 39: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/39.jpg)
039
◼ Problem
◼ Point-guided scene parsing
◼ Point-based distance metric learning
◼ Exploit semantic relationship across images
◼ Experimental results
◼ Good performance both quantitatively and
qualitatively
Conclusion
![Page 40: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/40.jpg)
Conclusion
![Page 41: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/41.jpg)
Weakly Supervised Learning for Real-World Computer Vision Applications & The 1st Learning from Imperfect Data (LID) Challenge
CVPR 2019 Workshop, Long Beach, CA
Object Segmentation on ILSVRC DET (Image-level Supervision) Scene Parsing on ADE20K (Point Supervision)
Task 1 Task 2
https://lidchallenge.github.io/
![Page 42: Towards Weakly Supervised Object Segmentation & Scene Parsing · Self-Erasing Network for Integral Object Attention Qibin Hou 1, Peng-Tao Jiang , Yunchao Wei 2, Ming-Ming Cheng1 1College](https://reader033.vdocuments.us/reader033/viewer/2022042222/5ec8fe5c4cbfa7299d5182a2/html5/thumbnails/42.jpg)
42 Thank you