agile lab rai - bnova...smart(er) content reccomendation thanks for contacts or questions:...

24
Agile Lab – RAI A NEW KIND OF VIDEO OBJECT RECOGNITION SYSTEM Dr. Alberto MESSINA – Researcher @CRIT-RAI Alberto Firpo – CEO @Agile Lab

Upload: others

Post on 10-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Agile Lab – RAIA NEW KIND OF VIDEO OBJECT

RECOGNITION SYSTEM

Dr. Alberto MESSINA – Researcher @CRIT-RAI

Alberto Firpo – CEO @Agile Lab

Page 2: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Agile Lab - What we do

SCALABLE TECHNOLOGIES

MACHINE LEARNING AND ARTIFICIAL INTELLIGENCE

Page 3: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Our background in image/video processing

Computer Vision “Mapping” Algorithms+ Deterministic Matching

(es.: SIFT)

Computer Vision “Mapping” Algorithms+ Deep Learning Decision

(es.: Selective Search + CNN)

Fully Deep Learning based(es.: CNN, fast, SSD, YOLO)

More FlexibilityMore PerformanceMore Use Cases

Page 4: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Performance drivers and trade off• Near Real Time vs Batch processing

• Frame frequency• Image Spatial Scaling

• Mean Average Precision

Page 5: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Some use cases – Deep Sat

• Precision: 0.98• Recall: 0.91• MAP: 0.78

Page 6: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Deep Sat: true positive

Page 7: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Deep Sat: true negative

Page 8: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Some use case - Deep Logo

Precision and recall per class:

Bmw: P: 0.97 R: 0.91Ferrari: P: 0.98 R: 0.85Ford: P: 1 R: 0.81HP: P: 1 R: 0.88nVidia: P: 1 R: 0.83Rolex: P: 0.97 R: 0.81Shell: P: 0.98 R: 0.80Apple: P: 1 R: 0.93

Page 9: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Online services today

Page 10: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint
Page 11: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint
Page 12: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Too much content …

Page 13: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

METADATA

Page 14: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

How to achieve automated contentunderstanding*?

* at reasonable costs …

Page 15: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

"Gina Lollobrigida sitting in a dark room"

All rights reserved – RAI Radiotelevisione Italiana

"faces": [{"age": 30,"gender": "Female","faceRectangle": {

"left": 352,"top": 186,"width": 55,"height": 55

}},{"age": 5,"gender": "Male","faceRectangle": {

"left": 436,"top": 247,"width": 47,"height": 47

}}

]

"tags": ["person","sitting","photo","woman","posing","front","holding","young","black","man","camera","white","girl","dark","standing","wearing","table","room","phone","living","computer","shirt"

]

Page 16: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

shot at Santa Maria Antiqua

Page 17: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

17

Why?

• Discovery of latent information throughvisual cues

• Impossible to annotate content from allpossible point of views

• Discovery of unexpected correlations• Linking and enriching

• Enabling user interaction in smartenvironments

• Tourism, culture

Page 18: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

18

AgileRAIarchitecture

How?

Page 19: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

19

AgileRAI video processing pipeline

MPEG CDVS

Extractor

CDVS

MatchingImage

DatabaseM

atc

hin

g

labels

CDVS Descriptors

Monitoring

Labels + timestampsSemantic

annotationInput video

(RAI1_YYYY-MM-DDThh:mm:ss.s,

Rialto_Bridge)

(Rialto_Bridge)

Page 20: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Semantic annotation pipeline

20

Video processing pipeline

Semanticenrichment

Labels +

timestamps

LOD repository

Triplestore‘Detection’ ontology

(RAI1_YYYY-MM-DDThh:mm:ss.s,

Rialto_Bridge)(dbpedia:Rialto_Bridge)

(owl:sameAs http://dbpedia.org/resource/Rialto_Bridge)

Page 21: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

UC1: video

browsing

21

Page 22: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

UC2: dynamic semantic tagging

22

Page 23: Agile Lab RAI - BNova...Smart(er) content reccomendation Thanks For contacts or questions: alberto.firpo@agilelab.it alberto.messina@rai.it Title Presentazione standard di PowerPoint

Efficient delivery is good but not enough

Content understandingand organisation is the key

Understanding, linking, enriching

Smart(er) content reccomendation