SlideShare ist ein Scribd-Unternehmen logo
1 von 38
Big models without big data:
Using deep networks for computer vision in
data-scarce settings
Jon Almazan, Cesar de Souza, Yohann Cabon,
Diane Larlus, Naila Murray, Jerome Revaud
Naver Labs Contributors
Yohann Cabon
Jerome Revaud
Cesar de Souza
Diane Larlus
Jon Almazan
Naila Murray
Deep learning for computer vision:
The data-scarcity challenge
Supervised deep learning :
J State-of-the-art for many CV tasks
L Requires lots of annotated data
Visual data is cheap and plentiful
Annotated data may be:
• Expensive
• Proprietary
• Non-feasible
How to use deep learning in data-scarce settings?
3
24 hrs of Photographyby Erik Kessels
Dealing with data-scarcity
4
Data synthesis
Domain adaptation
Data cleaning
Dealing with data-scarcity
5
Data synthesis
Domain adaptation
Data cleaning
Domain Adaptation
Leveraging annotated data in one or more related source
domains, to learn a model for unseen data in a target domain
Ground truth Prediction by PDP
Context: Attention prediction
7
Task: predict topographical attention map
Existing approaches: model it as a classification or regression task
Our approach: model attention as a stochastic process, using
probability distribution prediction (PDP)
Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
Approach
Model attention map as a generalized Bernoulli distribution
Apply novel loss functions that penalize distance btw. predicted(p) and target(t) distributions
Use fully-convolutional architecture for probability distribution prediction
8
Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
Data
Ground-truth attention data:
• Normally collected with eye-trackers
• Very expensive to collect
Jiang et al.*:
• introduce SALICON dataset
• use mouse-tracking as proxy:
We train our models with SALICON and fine-tune/test on
eye-tracking data
9
*Jiang et al. SALICON: Saliency in Context. CVPR 2015.
University of Kent
Results
10
Convergence of AUC using different loss functions Performance on SALICON test set
Results in source domain: mouse-tracking prediction
Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
Results
11
OSIE dataset
VOCA 2012 dataset
Results in target domain:
task-free eye-tracking prediction
Results in target domain:
task-dependent eye-tracking prediction
Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
Conclusion
12
Problem: attention map prediction
using limited target data
Solution: training with appropriate loss
functions, and pre-training with proxy
data
Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
Dealing with data-scarcity
13
Data synthesis
Domain adaptation
Data cleaning
Context: Instance-level Retrieval
Principle: Given a query image, find similar images in a (large)
database
14
Recent approaches
Recent methods leverage deep learning:
J Representations are compact and fast at test time!
Use standard networks designed for image classification:
L Not designed for retrieval
L Results significantly below the state-of-the-art
15
Can we learn to represent images for
retrieval?
Yes, if:
1. Training data is available
2. The network architecture can capture fine details
3. Training focuses on retrieval
16Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016.
Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
Obtaining Training Data
Public dataset of landmark images
• ~200K images
• 600 different landmarks (Eiffel tower, Rome colosseum, Big Ben…)
• Extremely noisy. Learning fails without clean data.
17
[Babenko et al, Neural codes @ ECCV14]
Prototypical view
Non-prototypical view
Wrong category
Obtaining Training Data
We proposed an automatic cleaning technique:
• Create graph per class using image matching
• Prune edges corresponding to low matching scores
• Use verified keypoint matches to mine bounding boxes
18
Public dataset of landmark images
• ~200K images
• 600 different landmarks (Eiffel tower, Rome colosseum, Big Ben…)
• Extremely noisy. Learning fails without clean data.
Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016.
Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
Obtaining Training Data
We proposed an automatic cleaning technique, resulting in:
• 40K spatially verified images
• Approximate bounding box annotations
• A new cleaned dataset, now publicly available
19
Public dataset of landmark images
• ~200K images
• 600 different landmarks (Eiffel tower, Rome colosseum, Big Ben…)
• Extremely noisy. Learning fails without clean data.
Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016.
Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
Proposed approach
Learning to rank images:
We propose a new three-stream Siamese Network: a network designed for
retrieval
20Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016.
Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
Experimental evaluation on standard
benchmarks
Oxford dataset
• 5k images
• 5k images + 100k distractor images
Paris dataset
• 6k images
INRIA Holidays dataset
• 1491 images
21
Experiments: Oxford 5k and Oxford 105k
Xerox Confidential 22
Deep Traditional Ours Deep Traditional Ours
82.7
84.3 84.9
86.9
89.4
50
60
70
80
90
100
MeanAveragePrecision
Oxford 5k
55.7
53.1
71.6 72.2
77.3
85
82.7
84.3 84.9
86.9
89.4
50
60
70
80
90
100
MeanAveragePrecision
Oxford 5k
76.7
80.2 79.5
85.3 84
45
50
55
60
65
70
75
80
85
90
95
100
MeanAveragePrecision
Oxford 105K
52.3
50.1
67.8
73.2
81.8
76.7
80.2 79.5
85.3 84
45
50
55
60
65
70
75
80
85
90
95
100
MeanAveragePrecision
Oxford 105K
52.3
50.1
67.8
73.2
81.8
76.7
80.2 79.5
85.3 84
93.6
45
50
55
60
65
70
75
80
85
90
95
100
MeanAveragePrecision
Oxford 105K
55.7
53.1
71.6 72.2
77.3
85
82.7
84.3 84.9
86.9
89.4
94.7
50
60
70
80
90
100
MeanAveragePrecision
Oxford 5k
Experiments: Paris 6k and INRIA Holidays
Xerox Confidential 23
Deep Traditional Ours Deep Traditional Ours
79.7
85.5
86.5 86.5
80.5
83.4
82.4
85.1
82.8
96.7
60
65
70
75
80
85
90
95
100
MeanAveragePrecision
Paris 6K
78.9
82
87.5
84.9
82.5
84.7
75.8
81.3
94.8
70
75
80
85
90
95
100
MeanAveragePrecision
INRIA Holidays
Qualitative results
24Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016.
Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
Conclusion
25
Problem: efficient instance-level image retrieval using deep networks
Solution: training with reliable annotations and an appropriate model architecture
Query
Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016.
Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
Dealing with data-scarcity
26
Data synthesis
Domain adaptation
Data cleaning
Synthetic Data for Computer Vision
Benefits
• Complete control
• Automatic annotations
• Quantity & variability
Challenges
• Chicken & egg problem?
• Technically feasible and cost-effective?
Our solution
• Off-the-shelf game engine (Unity)
• Seeding virtual worlds with limited real-world sensor data
• Automatic generation of all labels via shader programming
27
28
Gaidon et al. Virtual Worlds as Proxy
for Multi-Object Tracking Analysis.
CVPR 2016
Ros et al. The synthia dataset: A large collection of synthetic images
for semantic segmentation of urban scenes. CVPR 2016
Richter et al. Playing for Data: Ground Truth from
Computer Games. ECCV 2016
Synthetic Data for Computer Vision
Virtual worlds for action classification
From modelling vehicles to modelling human actions:
Orders of magnitude increase in complexity:
• non-rigid motion
• complex interactions with objects and people
• large diversity in viewpoints and appearance
How to create diverse, realistic, and physically-plausible
training videos?
Our solution: Procedural Human Action Videos (PHAV):
• generative model of human action videos
29
de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
30
Virtual worlds for action classification
Procedural Human Action Videos
PHAV Data modalities:
• RGB
• Depth
• Semantic Segmentation
• Instance Segmentation
• Horizontal Flow
• Vertical Flow
Extracted using Multiple Render Targets
31
32
Virtual worlds for action classification
de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
33
Adding PHAV helps training, particularly when real-world data is limited:
Naver Labs
Virtual worlds for action classification
de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
Conclusion
34
Problem: generate large-scale annotated synthetic videos useful for CV
Solution: modern game engine, real to virtual cloning, shaders
de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
Dealing with data-scarcity
35
Data synthesis
Domain adaptation
Data cleaning
Q & A
Thank you
Some numbers
Time to train the network: ~1 week on a single M40 GPU
Time to encode images: ~10 images per second on an M40 GPU
Total size per encoded image: 8Kb (128 images per Mb; dim=2048)
Time to compare images: millions of comparisons per second
• After PQ compression: 256 bytes/image with minor decrease in accuracy
Training memory requirements: ~3 x 7Gb
• 3-stream residual networks do not naively fit in memory!
• Each stream is processed sequentially: only one stream active at a time
38

Weitere ähnliche Inhalte

Was ist angesagt?

Multi-Tenant Storm Service on Hadoop Grid
Multi-Tenant Storm Service on Hadoop GridMulti-Tenant Storm Service on Hadoop Grid
Multi-Tenant Storm Service on Hadoop GridDataWorks Summit
 
Scaling Apache Storm (Hadoop Summit 2015)
Scaling Apache Storm (Hadoop Summit 2015)Scaling Apache Storm (Hadoop Summit 2015)
Scaling Apache Storm (Hadoop Summit 2015)Robert Evans
 
Apache Storm Internals
Apache Storm InternalsApache Storm Internals
Apache Storm InternalsHumoyun Ahmedov
 
[246]reasoning, attention and memory toward differentiable reasoning machines
[246]reasoning, attention and memory   toward differentiable reasoning machines[246]reasoning, attention and memory   toward differentiable reasoning machines
[246]reasoning, attention and memory toward differentiable reasoning machinesNAVER D2
 
[253] apache ni fi
[253] apache ni fi[253] apache ni fi
[253] apache ni fiNAVER D2
 
Clojure at BackType
Clojure at BackTypeClojure at BackType
Clojure at BackTypenathanmarz
 
Hadoop Summit Europe 2014: Apache Storm Architecture
Hadoop Summit Europe 2014: Apache Storm ArchitectureHadoop Summit Europe 2014: Apache Storm Architecture
Hadoop Summit Europe 2014: Apache Storm ArchitectureP. Taylor Goetz
 
Nervana and the Future of Computing
Nervana and the Future of ComputingNervana and the Future of Computing
Nervana and the Future of ComputingIntel Nervana
 
Tensorflow vs MxNet
Tensorflow vs MxNetTensorflow vs MxNet
Tensorflow vs MxNetAshish Bansal
 
Analysis big data by use php with storm
Analysis big data by use php with stormAnalysis big data by use php with storm
Analysis big data by use php with storm毅 吕
 
Introduction to Storm
Introduction to StormIntroduction to Storm
Introduction to StormEugene Dvorkin
 
Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16
Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16
Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16MLconf
 
Storm 2012-03-29
Storm 2012-03-29Storm 2012-03-29
Storm 2012-03-29Ted Dunning
 
Startup.Ml: Using neon for NLP and Localization Applications
Startup.Ml: Using neon for NLP and Localization Applications Startup.Ml: Using neon for NLP and Localization Applications
Startup.Ml: Using neon for NLP and Localization Applications Intel Nervana
 
Urs KĂśster - Convolutional and Recurrent Neural Networks
Urs KĂśster - Convolutional and Recurrent Neural NetworksUrs KĂśster - Convolutional and Recurrent Neural Networks
Urs KĂśster - Convolutional and Recurrent Neural NetworksIntel Nervana
 
Real time big data analytics with Storm by Ron Bodkin of Think Big Analytics
Real time big data analytics with Storm by Ron Bodkin of Think Big AnalyticsReal time big data analytics with Storm by Ron Bodkin of Think Big Analytics
Real time big data analytics with Storm by Ron Bodkin of Think Big AnalyticsData Con LA
 
Storm-on-YARN: Convergence of Low-Latency and Big-Data
Storm-on-YARN: Convergence of Low-Latency and Big-DataStorm-on-YARN: Convergence of Low-Latency and Big-Data
Storm-on-YARN: Convergence of Low-Latency and Big-DataDataWorks Summit
 
Real-time Big Data Processing with Storm
Real-time Big Data Processing with StormReal-time Big Data Processing with Storm
Real-time Big Data Processing with Stormviirya
 
Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016
Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016
Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016MLconf
 

Was ist angesagt? (20)

Multi-Tenant Storm Service on Hadoop Grid
Multi-Tenant Storm Service on Hadoop GridMulti-Tenant Storm Service on Hadoop Grid
Multi-Tenant Storm Service on Hadoop Grid
 
Scaling Apache Storm (Hadoop Summit 2015)
Scaling Apache Storm (Hadoop Summit 2015)Scaling Apache Storm (Hadoop Summit 2015)
Scaling Apache Storm (Hadoop Summit 2015)
 
Apache Storm Internals
Apache Storm InternalsApache Storm Internals
Apache Storm Internals
 
[246]reasoning, attention and memory toward differentiable reasoning machines
[246]reasoning, attention and memory   toward differentiable reasoning machines[246]reasoning, attention and memory   toward differentiable reasoning machines
[246]reasoning, attention and memory toward differentiable reasoning machines
 
[253] apache ni fi
[253] apache ni fi[253] apache ni fi
[253] apache ni fi
 
Resource Aware Scheduling in Apache Storm
Resource Aware Scheduling in Apache StormResource Aware Scheduling in Apache Storm
Resource Aware Scheduling in Apache Storm
 
Clojure at BackType
Clojure at BackTypeClojure at BackType
Clojure at BackType
 
Hadoop Summit Europe 2014: Apache Storm Architecture
Hadoop Summit Europe 2014: Apache Storm ArchitectureHadoop Summit Europe 2014: Apache Storm Architecture
Hadoop Summit Europe 2014: Apache Storm Architecture
 
Nervana and the Future of Computing
Nervana and the Future of ComputingNervana and the Future of Computing
Nervana and the Future of Computing
 
Tensorflow vs MxNet
Tensorflow vs MxNetTensorflow vs MxNet
Tensorflow vs MxNet
 
Analysis big data by use php with storm
Analysis big data by use php with stormAnalysis big data by use php with storm
Analysis big data by use php with storm
 
Introduction to Storm
Introduction to StormIntroduction to Storm
Introduction to Storm
 
Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16
Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16
Dr. Ike Nassi, Founder, TidalScale at MLconf NYC - 4/15/16
 
Storm 2012-03-29
Storm 2012-03-29Storm 2012-03-29
Storm 2012-03-29
 
Startup.Ml: Using neon for NLP and Localization Applications
Startup.Ml: Using neon for NLP and Localization Applications Startup.Ml: Using neon for NLP and Localization Applications
Startup.Ml: Using neon for NLP and Localization Applications
 
Urs KĂśster - Convolutional and Recurrent Neural Networks
Urs KĂśster - Convolutional and Recurrent Neural NetworksUrs KĂśster - Convolutional and Recurrent Neural Networks
Urs KĂśster - Convolutional and Recurrent Neural Networks
 
Real time big data analytics with Storm by Ron Bodkin of Think Big Analytics
Real time big data analytics with Storm by Ron Bodkin of Think Big AnalyticsReal time big data analytics with Storm by Ron Bodkin of Think Big Analytics
Real time big data analytics with Storm by Ron Bodkin of Think Big Analytics
 
Storm-on-YARN: Convergence of Low-Latency and Big-Data
Storm-on-YARN: Convergence of Low-Latency and Big-DataStorm-on-YARN: Convergence of Low-Latency and Big-Data
Storm-on-YARN: Convergence of Low-Latency and Big-Data
 
Real-time Big Data Processing with Storm
Real-time Big Data Processing with StormReal-time Big Data Processing with Storm
Real-time Big Data Processing with Storm
 
Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016
Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016
Alex Smola, Director of Machine Learning, AWS/Amazon, at MLconf SF 2016
 

Andere mochten auch

[221]똑똑한 인공지능 dj 비서 clova music
[221]똑똑한 인공지능 dj 비서 clova music[221]똑똑한 인공지능 dj 비서 clova music
[221]똑똑한 인공지능 dj 비서 clova musicNAVER D2
 
유연하고 확장성 있는 빅데이터 처리
유연하고 확장성 있는 빅데이터 처리유연하고 확장성 있는 빅데이터 처리
유연하고 확장성 있는 빅데이터 처리NAVER D2
 
[216]네이버 검색 사용자를 만족시켜라! 의도파악과 의미검색
[216]네이버 검색 사용자를 만족시켜라!   의도파악과 의미검색[216]네이버 검색 사용자를 만족시켜라!   의도파악과 의미검색
[216]네이버 검색 사용자를 만족시켜라! 의도파악과 의미검색NAVER D2
 
[241]large scale search with polysemous codes
[241]large scale search with polysemous codes[241]large scale search with polysemous codes
[241]large scale search with polysemous codesNAVER D2
 
[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms
[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms
[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbmsNAVER D2
 
[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼
[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼
[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼NAVER D2
 
[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화
[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화
[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화NAVER D2
 
[215]streetwise machine learning for painless parking
[215]streetwise machine learning for painless parking[215]streetwise machine learning for painless parking
[215]streetwise machine learning for painless parkingNAVER D2
 
[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법
[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법
[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법NAVER D2
 
[242]open stack neutron dataplane 구현
[242]open stack neutron   dataplane 구현[242]open stack neutron   dataplane 구현
[242]open stack neutron dataplane 구현NAVER D2
 
[234]멀티테넌트 하둡 클러스터 운영 경험기
[234]멀티테넌트 하둡 클러스터 운영 경험기[234]멀티테넌트 하둡 클러스터 운영 경험기
[234]멀티테넌트 하둡 클러스터 운영 경험기NAVER D2
 
[213]building ai to recreate our visual world
[213]building ai to recreate our visual world[213]building ai to recreate our visual world
[213]building ai to recreate our visual worldNAVER D2
 
[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기
[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기
[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기NAVER D2
 
[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기
[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기
[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기NAVER D2
 
인공지능추천시스템 airs개발기_모델링과시스템
인공지능추천시스템 airs개발기_모델링과시스템인공지능추천시스템 airs개발기_모델링과시스템
인공지능추천시스템 airs개발기_모델링과시스템NAVER D2
 
[232]mist 고성능 iot 스트림 처리 시스템
[232]mist 고성능 iot 스트림 처리 시스템[232]mist 고성능 iot 스트림 처리 시스템
[232]mist 고성능 iot 스트림 처리 시스템NAVER D2
 
백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스
백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스
백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스NAVER D2
 
[211] HBase 기반 검색 데이터 저장소 (공개용)
[211] HBase 기반 검색 데이터 저장소 (공개용)[211] HBase 기반 검색 데이터 저장소 (공개용)
[211] HBase 기반 검색 데이터 저장소 (공개용)NAVER D2
 
[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술
[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술
[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술NAVER D2
 
[141]네이버랩스의 로보틱스 연구 소개
[141]네이버랩스의 로보틱스 연구 소개[141]네이버랩스의 로보틱스 연구 소개
[141]네이버랩스의 로보틱스 연구 소개NAVER D2
 

Andere mochten auch (20)

[221]똑똑한 인공지능 dj 비서 clova music
[221]똑똑한 인공지능 dj 비서 clova music[221]똑똑한 인공지능 dj 비서 clova music
[221]똑똑한 인공지능 dj 비서 clova music
 
유연하고 확장성 있는 빅데이터 처리
유연하고 확장성 있는 빅데이터 처리유연하고 확장성 있는 빅데이터 처리
유연하고 확장성 있는 빅데이터 처리
 
[216]네이버 검색 사용자를 만족시켜라! 의도파악과 의미검색
[216]네이버 검색 사용자를 만족시켜라!   의도파악과 의미검색[216]네이버 검색 사용자를 만족시켜라!   의도파악과 의미검색
[216]네이버 검색 사용자를 만족시켜라! 의도파악과 의미검색
 
[241]large scale search with polysemous codes
[241]large scale search with polysemous codes[241]large scale search with polysemous codes
[241]large scale search with polysemous codes
 
[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms
[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms
[223]rye, 샤딩을 지원하는 오픈소스 관계형 dbms
 
[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼
[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼
[224]nsml 상상하는 모든 것이 이루어지는 클라우드 머신러닝 플랫폼
 
[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화
[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화
[231]운영체제 수준에서의 데이터베이스 성능 분석과 최적화
 
[215]streetwise machine learning for painless parking
[215]streetwise machine learning for painless parking[215]streetwise machine learning for painless parking
[215]streetwise machine learning for painless parking
 
[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법
[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법
[222]neural machine translation (nmt) 동작의 시각화 및 분석 방법
 
[242]open stack neutron dataplane 구현
[242]open stack neutron   dataplane 구현[242]open stack neutron   dataplane 구현
[242]open stack neutron dataplane 구현
 
[234]멀티테넌트 하둡 클러스터 운영 경험기
[234]멀티테넌트 하둡 클러스터 운영 경험기[234]멀티테넌트 하둡 클러스터 운영 경험기
[234]멀티테넌트 하둡 클러스터 운영 경험기
 
[213]building ai to recreate our visual world
[213]building ai to recreate our visual world[213]building ai to recreate our visual world
[213]building ai to recreate our visual world
 
[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기
[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기
[225]빅데이터를 위한 분산 딥러닝 플랫폼 만들기
 
[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기
[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기
[213] 의료 ai를 위해 세상에 없는 양질의 data 만드는 도구 제작하기
 
인공지능추천시스템 airs개발기_모델링과시스템
인공지능추천시스템 airs개발기_모델링과시스템인공지능추천시스템 airs개발기_모델링과시스템
인공지능추천시스템 airs개발기_모델링과시스템
 
[232]mist 고성능 iot 스트림 처리 시스템
[232]mist 고성능 iot 스트림 처리 시스템[232]mist 고성능 iot 스트림 처리 시스템
[232]mist 고성능 iot 스트림 처리 시스템
 
백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스
백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스
백억개의 로그를 모아 검색하고 분석하고 학습도 시켜보자 : 로기스
 
[211] HBase 기반 검색 데이터 저장소 (공개용)
[211] HBase 기반 검색 데이터 저장소 (공개용)[211] HBase 기반 검색 데이터 저장소 (공개용)
[211] HBase 기반 검색 데이터 저장소 (공개용)
 
[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술
[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술
[244]네트워크 모니터링 시스템(nms)을 지탱하는 기술
 
[141]네이버랩스의 로보틱스 연구 소개
[141]네이버랩스의 로보틱스 연구 소개[141]네이버랩스의 로보틱스 연구 소개
[141]네이버랩스의 로보틱스 연구 소개
 

Ähnlich wie [212]big models without big data using domain specific deep networks in data-scarce settings

深度學習在AOI的應用
深度學習在AOI的應用深度學習在AOI的應用
深度學習在AOI的應用CHENHuiMei
 
Introduction talk to Computer Vision
Introduction talk to Computer Vision Introduction talk to Computer Vision
Introduction talk to Computer Vision Chen Sagiv
 
Perception and Quality of Immersive Media
Perception and Quality of Immersive MediaPerception and Quality of Immersive Media
Perception and Quality of Immersive MediaAlpen-Adria-Universität
 
Report face recognition : ArganRecogn
Report face recognition :  ArganRecognReport face recognition :  ArganRecogn
Report face recognition : ArganRecognIlyas CHAOUA
 
"Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr...
"Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr..."Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr...
"Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr...Edge AI and Vision Alliance
 
The 'Rubble of the North' -a solution for modelling the irregular architectur...
The 'Rubble of the North' -a solution for modelling the irregular architectur...The 'Rubble of the North' -a solution for modelling the irregular architectur...
The 'Rubble of the North' -a solution for modelling the irregular architectur...3D ICONS Project
 
Structured Forests for Fast Edge Detection [Paper Presentation]
Structured Forests for Fast Edge Detection [Paper Presentation]Structured Forests for Fast Edge Detection [Paper Presentation]
Structured Forests for Fast Edge Detection [Paper Presentation]Mohammad Shaker
 
Scalable image recognition model with deep embedding
Scalable image recognition model with deep embeddingScalable image recognition model with deep embedding
Scalable image recognition model with deep embedding捷恩 蔡
 
Matchbox tool. Quality control for digital collections – SCAPE Training event...
Matchbox tool. Quality control for digital collections – SCAPE Training event...Matchbox tool. Quality control for digital collections – SCAPE Training event...
Matchbox tool. Quality control for digital collections – SCAPE Training event...SCAPE Project
 
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...IEEEFINALYEARSTUDENTPROJECT
 
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...IEEEMEMTECHSTUDENTSPROJECTS
 
IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...
IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...
IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...IEEEFINALYEARSTUDENTPROJECTS
 
Surveillance scene classification using machine learning
Surveillance scene classification using machine learningSurveillance scene classification using machine learning
Surveillance scene classification using machine learningUtkarsh Contractor
 
Scalable Deep Learning in ExtremeEarth-phiweek19
Scalable Deep Learning in ExtremeEarth-phiweek19Scalable Deep Learning in ExtremeEarth-phiweek19
Scalable Deep Learning in ExtremeEarth-phiweek19ExtremeEarth
 
Visual geometry with deep learning
Visual geometry with deep learningVisual geometry with deep learning
Visual geometry with deep learningNAVER Engineering
 
"Approaches for Vision-based Driver Monitoring," a Presentation from PathPart...
"Approaches for Vision-based Driver Monitoring," a Presentation from PathPart..."Approaches for Vision-based Driver Monitoring," a Presentation from PathPart...
"Approaches for Vision-based Driver Monitoring," a Presentation from PathPart...Edge AI and Vision Alliance
 
6 large-scale-learning.pptx
6 large-scale-learning.pptx6 large-scale-learning.pptx
6 large-scale-learning.pptxmustafa sarac
 
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)Universitat Politècnica de Catalunya
 

Ähnlich wie [212]big models without big data using domain specific deep networks in data-scarce settings (20)

深度學習在AOI的應用
深度學習在AOI的應用深度學習在AOI的應用
深度學習在AOI的應用
 
Introduction talk to Computer Vision
Introduction talk to Computer Vision Introduction talk to Computer Vision
Introduction talk to Computer Vision
 
Perception and Quality of Immersive Media
Perception and Quality of Immersive MediaPerception and Quality of Immersive Media
Perception and Quality of Immersive Media
 
Report face recognition : ArganRecogn
Report face recognition :  ArganRecognReport face recognition :  ArganRecogn
Report face recognition : ArganRecogn
 
Content-based Image Retrieval - Eva Mohedano - UPC Barcelona 2018
Content-based Image Retrieval - Eva Mohedano - UPC Barcelona 2018Content-based Image Retrieval - Eva Mohedano - UPC Barcelona 2018
Content-based Image Retrieval - Eva Mohedano - UPC Barcelona 2018
 
"Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr...
"Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr..."Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr...
"Enabling Ubiquitous Visual Intelligence Through Deep Learning," a Keynote Pr...
 
The 'Rubble of the North' -a solution for modelling the irregular architectur...
The 'Rubble of the North' -a solution for modelling the irregular architectur...The 'Rubble of the North' -a solution for modelling the irregular architectur...
The 'Rubble of the North' -a solution for modelling the irregular architectur...
 
Structured Forests for Fast Edge Detection [Paper Presentation]
Structured Forests for Fast Edge Detection [Paper Presentation]Structured Forests for Fast Edge Detection [Paper Presentation]
Structured Forests for Fast Edge Detection [Paper Presentation]
 
Learning where to look: focus and attention in deep vision
Learning where to look: focus and attention in deep visionLearning where to look: focus and attention in deep vision
Learning where to look: focus and attention in deep vision
 
Scalable image recognition model with deep embedding
Scalable image recognition model with deep embeddingScalable image recognition model with deep embedding
Scalable image recognition model with deep embedding
 
Matchbox tool. Quality control for digital collections – SCAPE Training event...
Matchbox tool. Quality control for digital collections – SCAPE Training event...Matchbox tool. Quality control for digital collections – SCAPE Training event...
Matchbox tool. Quality control for digital collections – SCAPE Training event...
 
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
 
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
2014 IEEE JAVA DATA MINING PROJECT Mining weakly labeled web facial images fo...
 
IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...
IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...
IEEE 2014 JAVA DATA MINING PROJECTS Mining weakly labeled web facial images f...
 
Surveillance scene classification using machine learning
Surveillance scene classification using machine learningSurveillance scene classification using machine learning
Surveillance scene classification using machine learning
 
Scalable Deep Learning in ExtremeEarth-phiweek19
Scalable Deep Learning in ExtremeEarth-phiweek19Scalable Deep Learning in ExtremeEarth-phiweek19
Scalable Deep Learning in ExtremeEarth-phiweek19
 
Visual geometry with deep learning
Visual geometry with deep learningVisual geometry with deep learning
Visual geometry with deep learning
 
"Approaches for Vision-based Driver Monitoring," a Presentation from PathPart...
"Approaches for Vision-based Driver Monitoring," a Presentation from PathPart..."Approaches for Vision-based Driver Monitoring," a Presentation from PathPart...
"Approaches for Vision-based Driver Monitoring," a Presentation from PathPart...
 
6 large-scale-learning.pptx
6 large-scale-learning.pptx6 large-scale-learning.pptx
6 large-scale-learning.pptx
 
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
 

Mehr von NAVER D2

[211] 인공지능이 인공지능 챗봇을 만든다
[211] 인공지능이 인공지능 챗봇을 만든다[211] 인공지능이 인공지능 챗봇을 만든다
[211] 인공지능이 인공지능 챗봇을 만든다NAVER D2
 
[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...
[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...
[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...NAVER D2
 
[215] Druid로 쉽고 빠르게 데이터 분석하기
[215] Druid로 쉽고 빠르게 데이터 분석하기[215] Druid로 쉽고 빠르게 데이터 분석하기
[215] Druid로 쉽고 빠르게 데이터 분석하기NAVER D2
 
[245]Papago Internals: 모델분석과 응용기술 개발
[245]Papago Internals: 모델분석과 응용기술 개발[245]Papago Internals: 모델분석과 응용기술 개발
[245]Papago Internals: 모델분석과 응용기술 개발NAVER D2
 
[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈
[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈
[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈NAVER D2
 
[235]Wikipedia-scale Q&A
[235]Wikipedia-scale Q&A[235]Wikipedia-scale Q&A
[235]Wikipedia-scale Q&ANAVER D2
 
[244]로봇이 현실 세계에 대해 학습하도록 만들기
[244]로봇이 현실 세계에 대해 학습하도록 만들기[244]로봇이 현실 세계에 대해 학습하도록 만들기
[244]로봇이 현실 세계에 대해 학습하도록 만들기NAVER D2
 
[243] Deep Learning to help student’s Deep Learning
[243] Deep Learning to help student’s Deep Learning[243] Deep Learning to help student’s Deep Learning
[243] Deep Learning to help student’s Deep LearningNAVER D2
 
[234]Fast & Accurate Data Annotation Pipeline for AI applications
[234]Fast & Accurate Data Annotation Pipeline for AI applications[234]Fast & Accurate Data Annotation Pipeline for AI applications
[234]Fast & Accurate Data Annotation Pipeline for AI applicationsNAVER D2
 
Old version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing
Old version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load BalancingOld version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing
Old version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load BalancingNAVER D2
 
[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지
[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지
[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지NAVER D2
 
[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기
[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기
[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기NAVER D2
 
[224]네이버 검색과 개인화
[224]네이버 검색과 개인화[224]네이버 검색과 개인화
[224]네이버 검색과 개인화NAVER D2
 
[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)
[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)
[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)NAVER D2
 
[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기
[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기
[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기NAVER D2
 
[213] Fashion Visual Search
[213] Fashion Visual Search[213] Fashion Visual Search
[213] Fashion Visual SearchNAVER D2
 
[232] TensorRT를 활용한 딥러닝 Inference 최적화
[232] TensorRT를 활용한 딥러닝 Inference 최적화[232] TensorRT를 활용한 딥러닝 Inference 최적화
[232] TensorRT를 활용한 딥러닝 Inference 최적화NAVER D2
 
[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지
[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지
[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지NAVER D2
 
[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터
[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터
[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터NAVER D2
 
[223]기계독해 QA: 검색인가, NLP인가?
[223]기계독해 QA: 검색인가, NLP인가?[223]기계독해 QA: 검색인가, NLP인가?
[223]기계독해 QA: 검색인가, NLP인가?NAVER D2
 

Mehr von NAVER D2 (20)

[211] 인공지능이 인공지능 챗봇을 만든다
[211] 인공지능이 인공지능 챗봇을 만든다[211] 인공지능이 인공지능 챗봇을 만든다
[211] 인공지능이 인공지능 챗봇을 만든다
 
[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...
[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...
[233] 대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing: Maglev Hashing Scheduler i...
 
[215] Druid로 쉽고 빠르게 데이터 분석하기
[215] Druid로 쉽고 빠르게 데이터 분석하기[215] Druid로 쉽고 빠르게 데이터 분석하기
[215] Druid로 쉽고 빠르게 데이터 분석하기
 
[245]Papago Internals: 모델분석과 응용기술 개발
[245]Papago Internals: 모델분석과 응용기술 개발[245]Papago Internals: 모델분석과 응용기술 개발
[245]Papago Internals: 모델분석과 응용기술 개발
 
[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈
[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈
[236] 스트림 저장소 최적화 이야기: 아파치 드루이드로부터 얻은 교훈
 
[235]Wikipedia-scale Q&A
[235]Wikipedia-scale Q&A[235]Wikipedia-scale Q&A
[235]Wikipedia-scale Q&A
 
[244]로봇이 현실 세계에 대해 학습하도록 만들기
[244]로봇이 현실 세계에 대해 학습하도록 만들기[244]로봇이 현실 세계에 대해 학습하도록 만들기
[244]로봇이 현실 세계에 대해 학습하도록 만들기
 
[243] Deep Learning to help student’s Deep Learning
[243] Deep Learning to help student’s Deep Learning[243] Deep Learning to help student’s Deep Learning
[243] Deep Learning to help student’s Deep Learning
 
[234]Fast & Accurate Data Annotation Pipeline for AI applications
[234]Fast & Accurate Data Annotation Pipeline for AI applications[234]Fast & Accurate Data Annotation Pipeline for AI applications
[234]Fast & Accurate Data Annotation Pipeline for AI applications
 
Old version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing
Old version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load BalancingOld version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing
Old version: [233]대형 컨테이너 클러스터에서의 고가용성 Network Load Balancing
 
[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지
[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지
[226]NAVER 광고 deep click prediction: 모델링부터 서빙까지
 
[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기
[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기
[225]NSML: 머신러닝 플랫폼 서비스하기 & 모델 튜닝 자동화하기
 
[224]네이버 검색과 개인화
[224]네이버 검색과 개인화[224]네이버 검색과 개인화
[224]네이버 검색과 개인화
 
[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)
[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)
[216]Search Reliability Engineering (부제: 지진에도 흔들리지 않는 네이버 검색시스템)
 
[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기
[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기
[214] Ai Serving Platform: 하루 수 억 건의 인퍼런스를 처리하기 위한 고군분투기
 
[213] Fashion Visual Search
[213] Fashion Visual Search[213] Fashion Visual Search
[213] Fashion Visual Search
 
[232] TensorRT를 활용한 딥러닝 Inference 최적화
[232] TensorRT를 활용한 딥러닝 Inference 최적화[232] TensorRT를 활용한 딥러닝 Inference 최적화
[232] TensorRT를 활용한 딥러닝 Inference 최적화
 
[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지
[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지
[242]컴퓨터 비전을 이용한 실내 지도 자동 업데이트 방법: 딥러닝을 통한 POI 변화 탐지
 
[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터
[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터
[212]C3, 데이터 처리에서 서빙까지 가능한 하둡 클러스터
 
[223]기계독해 QA: 검색인가, NLP인가?
[223]기계독해 QA: 검색인가, NLP인가?[223]기계독해 QA: 검색인가, NLP인가?
[223]기계독해 QA: 검색인가, NLP인가?
 

KĂźrzlich hochgeladen

Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024The Digital Insurer
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingEdi Saputra
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyKhushali Kathiriya
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024SynarionITSolutions
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfsudhanshuwaghmare1
 
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodPolkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodJuan lago vĂĄzquez
 
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Principled Technologies
 
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsTop 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsRoshan Dwivedi
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobeapidays
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century educationjfdjdjcjdnsjd
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businesspanagenda
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...Neo4j
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024The Digital Insurer
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherRemote DBA Services
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Scriptwesley chun
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...apidays
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Miguel AraĂşjo
 
Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...apidays
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonAnna Loughnan Colquhoun
 

KĂźrzlich hochgeladen (20)

Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : Uncertainty
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdf
 
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodPolkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
 
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
 
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsTop 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Script
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
 

[212]big models without big data using domain specific deep networks in data-scarce settings

  • 1. Big models without big data: Using deep networks for computer vision in data-scarce settings Jon Almazan, Cesar de Souza, Yohann Cabon, Diane Larlus, Naila Murray, Jerome Revaud
  • 2. Naver Labs Contributors Yohann Cabon Jerome Revaud Cesar de Souza Diane Larlus Jon Almazan Naila Murray
  • 3. Deep learning for computer vision: The data-scarcity challenge Supervised deep learning : J State-of-the-art for many CV tasks L Requires lots of annotated data Visual data is cheap and plentiful Annotated data may be: • Expensive • Proprietary • Non-feasible How to use deep learning in data-scarce settings? 3 24 hrs of Photographyby Erik Kessels
  • 4. Dealing with data-scarcity 4 Data synthesis Domain adaptation Data cleaning
  • 5. Dealing with data-scarcity 5 Data synthesis Domain adaptation Data cleaning
  • 6. Domain Adaptation Leveraging annotated data in one or more related source domains, to learn a model for unseen data in a target domain
  • 7. Ground truth Prediction by PDP Context: Attention prediction 7 Task: predict topographical attention map Existing approaches: model it as a classification or regression task Our approach: model attention as a stochastic process, using probability distribution prediction (PDP) Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
  • 8. Approach Model attention map as a generalized Bernoulli distribution Apply novel loss functions that penalize distance btw. predicted(p) and target(t) distributions Use fully-convolutional architecture for probability distribution prediction 8 Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
  • 9. Data Ground-truth attention data: • Normally collected with eye-trackers • Very expensive to collect Jiang et al.*: • introduce SALICON dataset • use mouse-tracking as proxy: We train our models with SALICON and fine-tune/test on eye-tracking data 9 *Jiang et al. SALICON: Saliency in Context. CVPR 2015. University of Kent
  • 10. Results 10 Convergence of AUC using different loss functions Performance on SALICON test set Results in source domain: mouse-tracking prediction Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
  • 11. Results 11 OSIE dataset VOCA 2012 dataset Results in target domain: task-free eye-tracking prediction Results in target domain: task-dependent eye-tracking prediction Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
  • 12. Conclusion 12 Problem: attention map prediction using limited target data Solution: training with appropriate loss functions, and pre-training with proxy data Jetley, Murray, Vig. End-to-End Saliency Mapping via Probability Distribution Prediction. CVPR 2016.
  • 13. Dealing with data-scarcity 13 Data synthesis Domain adaptation Data cleaning
  • 14. Context: Instance-level Retrieval Principle: Given a query image, find similar images in a (large) database 14
  • 15. Recent approaches Recent methods leverage deep learning: J Representations are compact and fast at test time! Use standard networks designed for image classification: L Not designed for retrieval L Results significantly below the state-of-the-art 15
  • 16. Can we learn to represent images for retrieval? Yes, if: 1. Training data is available 2. The network architecture can capture fine details 3. Training focuses on retrieval 16Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016. Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
  • 17. Obtaining Training Data Public dataset of landmark images • ~200K images • 600 different landmarks (Eiffel tower, Rome colosseum, Big Ben…) • Extremely noisy. Learning fails without clean data. 17 [Babenko et al, Neural codes @ ECCV14] Prototypical view Non-prototypical view Wrong category
  • 18. Obtaining Training Data We proposed an automatic cleaning technique: • Create graph per class using image matching • Prune edges corresponding to low matching scores • Use verified keypoint matches to mine bounding boxes 18 Public dataset of landmark images • ~200K images • 600 different landmarks (Eiffel tower, Rome colosseum, Big Ben…) • Extremely noisy. Learning fails without clean data. Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016. Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
  • 19. Obtaining Training Data We proposed an automatic cleaning technique, resulting in: • 40K spatially verified images • Approximate bounding box annotations • A new cleaned dataset, now publicly available 19 Public dataset of landmark images • ~200K images • 600 different landmarks (Eiffel tower, Rome colosseum, Big Ben…) • Extremely noisy. Learning fails without clean data. Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016. Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
  • 20. Proposed approach Learning to rank images: We propose a new three-stream Siamese Network: a network designed for retrieval 20Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016. Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
  • 21. Experimental evaluation on standard benchmarks Oxford dataset • 5k images • 5k images + 100k distractor images Paris dataset • 6k images INRIA Holidays dataset • 1491 images 21
  • 22. Experiments: Oxford 5k and Oxford 105k Xerox Confidential 22 Deep Traditional Ours Deep Traditional Ours 82.7 84.3 84.9 86.9 89.4 50 60 70 80 90 100 MeanAveragePrecision Oxford 5k 55.7 53.1 71.6 72.2 77.3 85 82.7 84.3 84.9 86.9 89.4 50 60 70 80 90 100 MeanAveragePrecision Oxford 5k 76.7 80.2 79.5 85.3 84 45 50 55 60 65 70 75 80 85 90 95 100 MeanAveragePrecision Oxford 105K 52.3 50.1 67.8 73.2 81.8 76.7 80.2 79.5 85.3 84 45 50 55 60 65 70 75 80 85 90 95 100 MeanAveragePrecision Oxford 105K 52.3 50.1 67.8 73.2 81.8 76.7 80.2 79.5 85.3 84 93.6 45 50 55 60 65 70 75 80 85 90 95 100 MeanAveragePrecision Oxford 105K 55.7 53.1 71.6 72.2 77.3 85 82.7 84.3 84.9 86.9 89.4 94.7 50 60 70 80 90 100 MeanAveragePrecision Oxford 5k
  • 23. Experiments: Paris 6k and INRIA Holidays Xerox Confidential 23 Deep Traditional Ours Deep Traditional Ours 79.7 85.5 86.5 86.5 80.5 83.4 82.4 85.1 82.8 96.7 60 65 70 75 80 85 90 95 100 MeanAveragePrecision Paris 6K 78.9 82 87.5 84.9 82.5 84.7 75.8 81.3 94.8 70 75 80 85 90 95 100 MeanAveragePrecision INRIA Holidays
  • 24. Qualitative results 24Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016. Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
  • 25. Conclusion 25 Problem: efficient instance-level image retrieval using deep networks Solution: training with reliable annotations and an appropriate model architecture Query Gordo, Almazan, Revaud, Larlus. Deep Image Retrieval: Learning global representations for image search. ECCV 2016. Gordo, Almazan, Revaud, Larlus. End-to-End Learning of Deep Visual Representations for Image Retrieval. IJCV 2017.
  • 26. Dealing with data-scarcity 26 Data synthesis Domain adaptation Data cleaning
  • 27. Synthetic Data for Computer Vision Benefits • Complete control • Automatic annotations • Quantity & variability Challenges • Chicken & egg problem? • Technically feasible and cost-effective? Our solution • Off-the-shelf game engine (Unity) • Seeding virtual worlds with limited real-world sensor data • Automatic generation of all labels via shader programming 27
  • 28. 28 Gaidon et al. Virtual Worlds as Proxy for Multi-Object Tracking Analysis. CVPR 2016 Ros et al. The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes. CVPR 2016 Richter et al. Playing for Data: Ground Truth from Computer Games. ECCV 2016 Synthetic Data for Computer Vision
  • 29. Virtual worlds for action classification From modelling vehicles to modelling human actions: Orders of magnitude increase in complexity: • non-rigid motion • complex interactions with objects and people • large diversity in viewpoints and appearance How to create diverse, realistic, and physically-plausible training videos? Our solution: Procedural Human Action Videos (PHAV): • generative model of human action videos 29 de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
  • 30. 30 Virtual worlds for action classification
  • 31. Procedural Human Action Videos PHAV Data modalities: • RGB • Depth • Semantic Segmentation • Instance Segmentation • Horizontal Flow • Vertical Flow Extracted using Multiple Render Targets 31
  • 32. 32 Virtual worlds for action classification de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
  • 33. 33 Adding PHAV helps training, particularly when real-world data is limited: Naver Labs Virtual worlds for action classification de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
  • 34. Conclusion 34 Problem: generate large-scale annotated synthetic videos useful for CV Solution: modern game engine, real to virtual cloning, shaders de Souza, Cabon, Gaidon, Lopez. Procedural Generation of Videos to Train Deep Action Recognition Networks. CVPR 2017.
  • 35. Dealing with data-scarcity 35 Data synthesis Domain adaptation Data cleaning
  • 36. Q & A
  • 38. Some numbers Time to train the network: ~1 week on a single M40 GPU Time to encode images: ~10 images per second on an M40 GPU Total size per encoded image: 8Kb (128 images per Mb; dim=2048) Time to compare images: millions of comparisons per second • After PQ compression: 256 bytes/image with minor decrease in accuracy Training memory requirements: ~3 x 7Gb • 3-stream residual networks do not naively fit in memory! • Each stream is processed sequentially: only one stream active at a time 38