SlideShare une entreprise Scribd logo
1  sur  22
Télécharger pour lire hors ligne
Real-Time Modelling Visual
Scenes with Biological
Inspiration
Kofi Appiah
Sheffield Hallam University
AI now and before
• Computer Vision and natural language processing have improved
significantly over the past 10 years.
• Image recognition and classification systems
• Apple photo organiser, Facebook face recognition.
• Robot use in warehouse
• Amazon warehouse robots (https://www.youtube.com/watch?v=4sEVX4mPuto)
• Medical image analysis for healthcare
• non-invasive diagnosis
• Agriculture, sport, manufacturing, autonomous cars technology.
• Crop yield, goal-line technology, defective products, people detection.
Human level face recognition Taigman et. al. CVPR2014
Why AI acceleration
• Better algorithms that learn from examples not predefined rules
• Deep learning
• Neural networks
• Machine perception
• Availability of data – Big Data
• Internet images, YouTube videos, Facebook images
• High Performance Computing
• Field Programmable Gate Arrays (FPGAs)
• Graphics Processing Units (GPU)
IEEE Spectrum
Key Achievements
• Visual recognition with high accuracies.
• 3D reconstruction of an environment
Mask R-CNN He et. al. ICCV2017
Litjens et. al. 2017
Johnson et. al. CVPR2015
Driverless cars - Mathworks
Faster R-CNN TPAMI 2017
Where things fall apart
• March 18, 2018, Uber’s autonomous car hit and killed 49-year-old as
she was walking her bike across the street.
• https://www.youtube.com/watch?v=7iTshCm41Ko
• Novel and imperfect system
• March 23 2018, autopilot Tesla slammed into concrete killing driver.
• Security robots attacking a kid in a shopping area, July 2016.
• Robot failure to open different doors – which training mode.
• Reinforced learning
• Supervised or Unsupervised?
Why things go wrong
• For autonomous cars, the state of the art is good and providing
bounding boxes of objects in the scene.
• What is missing is an interpretation of the scene.
• No contextual reasoning.
• Robot navigation
• Decision making might be optimal but not feasible or safe.
• Modelling in a crowded scene to infer interaction
• Modelling very unusual situations with little or no data
• Things that human are capable of, e.g. dealing with complex scenes
Fei-Fei Li
Unsupervised Background Subtraction
• Image Segmentation separate moving
objects from the background.
• Background subtraction is a practical
approach when the image sensor is
stationary.
• Background Modelling techniques
- Unimodal
- Multimodal
W4 and Grimson’s Algorithm – 2000s
• Requires manual initialization of
the Maximum (M), Minimum (m)
& inter-frame difference (D)
• Pixel x of image I is foreground if
|m(x)-It(x)|>D(x) or |M(x)-
It(x)|>D(x)
• Detection, Motion & change
history maps used for outdoor
scene.
• Use of fixed-point update values.
• Bimodal can’t model problems like
moving foliage and lighting
changes.
• Mixture of Gaussians with
associated weights to model each
pixel.
• Parameters are updated as follows:
• The first B distributions, ordered
by weight represents the
background
• Robust in modeling multimodal
background.
• Suffers from blending effect and
uses floating point in all updates
Efficient Hardware Implementation
• Maintains K clusters each with weight wk, central value ck
and implied global range [ck-15, ck+15]
• Weights and central values of all clusters are initialized 0,
and updated as follows:
• Uses both pixel and frame-level processing
• The first B distributions, ordered by weight represents the
background






+
=
−
−
otherwise
64
63
clustermatchingfor the
64
1
64
63
1,
1,
,
tk
tk
tk








+
=
−
−
otherwise
clustermatching
8
1
8
7
,,1,
,,,1,
,,,
jitk
jijitk
jitk
c
Xc
c






= =
b
k
ib TB
1
minarg 
Appiah et al FPT 2005
TULIPP – The game changer!
• Tools to help real-time computer vision developer to focus on:
• core application development by automating recurring, but critical,
tasks such as performance instrumentation
• Design space exploration and
• Vendor tool configuration.
• Making it possible for the designer to get the required
performance in speed, coupled with power constraints without
having to worry too much about the architecture.
Imaging before Deep Learning
Before
• Standard feature detectors
• SIFT, HOG, LBP
• Different algorithms for object
detection
• Requires small amount of data
• Useful for measurement and
labelling
After
• Featured are learnt and stacked
according to data
• Same algorithm that adapts to the
data
• Requires huge volume of data
• Useful for labelling
MathworksDalal & Triggs cc.gatech.edu
Deep CNN – Overview
• Uses convolution to preserve the spatial
structure of the input image
• Instead of a sigmoid activation function,
ReLU (rectified linear unit) is often used
• Encourages sparsity of synapses as
the value approaches zero (0).
Credit : Fei-Fei Li CS231n; Bala Amavasai – IEEE & M. Turner
Feature Maps - Several feature maps are used to identify various local features
• Several feature maps are used to
identify various local features.
• Each convolution filter can be tuned
to edges of different
• Orientation, Frequency, Phase, Colour, etc
• Capture some aspects of neural response
• But neural data not used in training
Sparse local connectivity
• For an input image of size 7x7
• The convolution filter 3x3
• The output image will be 5x5
• (Image – Filter )/stride + 1
• A sample filter for horizontal and
vertical gradient.
Way forward
• Computer Vision meets Cognitive Science and Neuroscience
Fei-Fei Li & Justin Johnson & Serena Yeung
The Challenge
• The success stories about the rise of Convolutional Neural Networks
(CNNs) capable of learning high-level features in object recognition
keeps increasing
• due to the availability of large datasets like ImageNet
• However, performance at scene recognition has not attained the same
level of success.
• Yet large scene databases like SUN and Places do exist
• Maybe the current deep features trained from ImageNet are not
competitive enough for such tasks.
• But do primates and humans actually do a raster scan to understand a
scene?
• CNNs fail to capture insensitivity to perturbations of an image
Possible Solution
• Performance accuracies in CNNs relies on a huge search space.
• The need for more biological guidance from the visual cortex
• Multi-disciplinary research in neuroscience, psychology,
physiology, shows that:
• object recognition in visual cortex is modulated via the ventral stream
• Neuronal signals from the retina are transformed into high-level
representation for object recognition.
• Computer Scientist working with neuroscientist, psychologist,
etc. would have better models for understanding scenes.
Reported Successes
• A biologically Inspired Deep CNN Model [Zhang et al. 2016]
• Simulates the V1, V2, V4 and IT layers of the human ventral stream
• Uses convolutional layers with varied sizes and complexities
• Increased concurrency for improved processing speed
• Outperformed seven other CNN techniques using four datasets.
• You Only Look Once (YOLOv2) [Redmon and Farhadi CVPR2017]
• Based on the assumption that humans glance at an image
• Does not rely on sliding window like other deep learning approaches
• Outperforms Deformable Part Models (DPM) and Regional CNN.
Scene understanding with DNN
• Learning Deep Features for Scene Recognition using Places
Database [Zhou et al. NIPS2014]
• Uses CNN to learn features from the scene
• Combined various local and global features to understand the scene
• Presents scene categories where machines perform like humans.
• Humans, but Not Deep Neural Networks, Often Miss Giant
Targets in Scenes [Eckstein et al. Current Biology 2017]
• Humans often miss unusual sized targets during visual search
• Deep learning does not exhibit such deficit with targets
• Is that a good thing or not?
Our motivation
• Missing giant targets is a functional brain strategy to discount
distractors
Eckstein et al. Current Biology 2017
Our Approach
• To understand how humans and primates recognise scenes
• Provide them with samples of indoor scenes
• Ask them to identify specific objects
• Observe their recall mechanism, if spatial relationship plays a role
• Model the scene to account for the experimental results
• Incorporate global and local descriptors
• Construct a relationship vector
Lunchroom image : PASSTA Dataset
Summary
• Computer vision and machine learning have improved over the
years, thanks to more data and processing power.
• Global scene understanding is still a challenge.
• Multi-disciplinary effort required to take computer vision to the
next level, acceptable for applications like driverless cars.
• We aim to combine positives of CNN with what humans are
good at for scene understanding.
• TULIPP offers the platform with toolchain to drive this agenda.

Contenu connexe

Tendances

Computer vision for interactive computer graphics
Computer vision for interactive computer graphicsComputer vision for interactive computer graphics
Computer vision for interactive computer graphicsShah Alam Sabuj
 
M.Sc. Thesis - Automatic People Counting in Crowded Scenes
M.Sc. Thesis - Automatic People Counting in Crowded ScenesM.Sc. Thesis - Automatic People Counting in Crowded Scenes
M.Sc. Thesis - Automatic People Counting in Crowded ScenesAhmed Gad
 
Architecture Design for Deep Neural Networks I
Architecture Design for Deep Neural Networks IArchitecture Design for Deep Neural Networks I
Architecture Design for Deep Neural Networks IWanjin Yu
 
IRJET- 3D Object Recognition of Car Image Detection
IRJET-  	  3D Object Recognition of Car Image DetectionIRJET-  	  3D Object Recognition of Car Image Detection
IRJET- 3D Object Recognition of Car Image DetectionIRJET Journal
 
Image Search: Then and Now
Image Search: Then and NowImage Search: Then and Now
Image Search: Then and NowSi Krishan
 
MOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNET
MOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNETMOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNET
MOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNETgerogepatton
 
Remote Sensing Image Scene Classification
Remote Sensing Image Scene ClassificationRemote Sensing Image Scene Classification
Remote Sensing Image Scene ClassificationGaurav Singh
 
Qualcomm research-imagenet2015
Qualcomm research-imagenet2015Qualcomm research-imagenet2015
Qualcomm research-imagenet2015Bilkent University
 
IRJET- Exploring Image Super Resolution Techniques
IRJET- Exploring Image Super Resolution TechniquesIRJET- Exploring Image Super Resolution Techniques
IRJET- Exploring Image Super Resolution TechniquesIRJET Journal
 
2019 cvpr paper_overview
2019 cvpr paper_overview2019 cvpr paper_overview
2019 cvpr paper_overviewLEE HOSEONG
 
Image recognition
Image recognitionImage recognition
Image recognitionJoel Jose
 
IEEE EED2021 AI use cases in Computer Vision
IEEE EED2021 AI use cases in Computer VisionIEEE EED2021 AI use cases in Computer Vision
IEEE EED2021 AI use cases in Computer VisionSAMeh Zaghloul
 
Weave-D - 2nd Progress Evaluation Presentation
Weave-D - 2nd Progress Evaluation PresentationWeave-D - 2nd Progress Evaluation Presentation
Weave-D - 2nd Progress Evaluation Presentationlasinducharith
 
PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...
PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...
PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...광희 이
 
Nicolae_Denut_Theses
Nicolae_Denut_ThesesNicolae_Denut_Theses
Nicolae_Denut_ThesesNicolae Denut
 
Neural Networks for Pattern Recognition
Neural Networks for Pattern RecognitionNeural Networks for Pattern Recognition
Neural Networks for Pattern RecognitionVipra Singh
 
Paper id 6122018109
Paper id 6122018109Paper id 6122018109
Paper id 6122018109IJRAT
 

Tendances (20)

Deep learning and computer vision
Deep learning and computer visionDeep learning and computer vision
Deep learning and computer vision
 
Computer vision for interactive computer graphics
Computer vision for interactive computer graphicsComputer vision for interactive computer graphics
Computer vision for interactive computer graphics
 
M.Sc. Thesis - Automatic People Counting in Crowded Scenes
M.Sc. Thesis - Automatic People Counting in Crowded ScenesM.Sc. Thesis - Automatic People Counting in Crowded Scenes
M.Sc. Thesis - Automatic People Counting in Crowded Scenes
 
Architecture Design for Deep Neural Networks I
Architecture Design for Deep Neural Networks IArchitecture Design for Deep Neural Networks I
Architecture Design for Deep Neural Networks I
 
IRJET- 3D Object Recognition of Car Image Detection
IRJET-  	  3D Object Recognition of Car Image DetectionIRJET-  	  3D Object Recognition of Car Image Detection
IRJET- 3D Object Recognition of Car Image Detection
 
Image Search: Then and Now
Image Search: Then and NowImage Search: Then and Now
Image Search: Then and Now
 
Fame cvpr
Fame cvprFame cvpr
Fame cvpr
 
MOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNET
MOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNETMOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNET
MOTION PREDICTION USING DEPTH INFORMATION OF HUMAN ARM BASED ON ALEXNET
 
Remote Sensing Image Scene Classification
Remote Sensing Image Scene ClassificationRemote Sensing Image Scene Classification
Remote Sensing Image Scene Classification
 
Qualcomm research-imagenet2015
Qualcomm research-imagenet2015Qualcomm research-imagenet2015
Qualcomm research-imagenet2015
 
Image recognition
Image recognitionImage recognition
Image recognition
 
IRJET- Exploring Image Super Resolution Techniques
IRJET- Exploring Image Super Resolution TechniquesIRJET- Exploring Image Super Resolution Techniques
IRJET- Exploring Image Super Resolution Techniques
 
2019 cvpr paper_overview
2019 cvpr paper_overview2019 cvpr paper_overview
2019 cvpr paper_overview
 
Image recognition
Image recognitionImage recognition
Image recognition
 
IEEE EED2021 AI use cases in Computer Vision
IEEE EED2021 AI use cases in Computer VisionIEEE EED2021 AI use cases in Computer Vision
IEEE EED2021 AI use cases in Computer Vision
 
Weave-D - 2nd Progress Evaluation Presentation
Weave-D - 2nd Progress Evaluation PresentationWeave-D - 2nd Progress Evaluation Presentation
Weave-D - 2nd Progress Evaluation Presentation
 
PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...
PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...
PR100: SeedNet: Automatic Seed Generation with Deep Reinforcement Learning fo...
 
Nicolae_Denut_Theses
Nicolae_Denut_ThesesNicolae_Denut_Theses
Nicolae_Denut_Theses
 
Neural Networks for Pattern Recognition
Neural Networks for Pattern RecognitionNeural Networks for Pattern Recognition
Neural Networks for Pattern Recognition
 
Paper id 6122018109
Paper id 6122018109Paper id 6122018109
Paper id 6122018109
 

Similaire à HiPEAC 2019 Workshop - Real-Time Modelling Visual Scenes with Biological Inspiration

Makine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine Learning
Makine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine LearningMakine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine Learning
Makine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine LearningAli Alkan
 
Introduction talk to Computer Vision
Introduction talk to Computer Vision Introduction talk to Computer Vision
Introduction talk to Computer Vision Chen Sagiv
 
Scene classification using Convolutional Neural Networks - Jayani Withanawasam
Scene classification using Convolutional Neural Networks - Jayani WithanawasamScene classification using Convolutional Neural Networks - Jayani Withanawasam
Scene classification using Convolutional Neural Networks - Jayani WithanawasamWithTheBest
 
Intelligent image processing
Intelligent image processingIntelligent image processing
Intelligent image processingAndrew Stewart
 
Detection and recognition of face using neural network
Detection and recognition of face using neural networkDetection and recognition of face using neural network
Detection and recognition of face using neural networkSmriti Tikoo
 
MDEC Data Matters Series: machine learning and Deep Learning, A Primer
MDEC Data Matters Series: machine learning and Deep Learning, A PrimerMDEC Data Matters Series: machine learning and Deep Learning, A Primer
MDEC Data Matters Series: machine learning and Deep Learning, A PrimerPoo Kuan Hoong
 
Biometric Recognition using Deep Learning
Biometric Recognition using Deep LearningBiometric Recognition using Deep Learning
Biometric Recognition using Deep LearningSahithiKotha2
 
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence Spracklen
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence SpracklenSBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence Spracklen
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence SpracklenNumenta
 
Real Time Object Dectection using machine learning
Real Time Object Dectection using machine learningReal Time Object Dectection using machine learning
Real Time Object Dectection using machine learningpratik pratyay
 
Image Segmentation Using Deep Learning : A survey
Image Segmentation Using Deep Learning : A surveyImage Segmentation Using Deep Learning : A survey
Image Segmentation Using Deep Learning : A surveyNUPUR YADAV
 
"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
 
TechnicalBackgroundOverview
TechnicalBackgroundOverviewTechnicalBackgroundOverview
TechnicalBackgroundOverviewMotaz El-Saban
 
Deep learning introduction
Deep learning introductionDeep learning introduction
Deep learning introductionAdwait Bhave
 
Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...
Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...
Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...gabrielesisinna
 

Similaire à HiPEAC 2019 Workshop - Real-Time Modelling Visual Scenes with Biological Inspiration (20)

Makine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine Learning
Makine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine LearningMakine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine Learning
Makine Öğrenmesi ile Görüntü Tanıma | Image Recognition using Machine Learning
 
Introduction talk to Computer Vision
Introduction talk to Computer Vision Introduction talk to Computer Vision
Introduction talk to Computer Vision
 
Scene classification using Convolutional Neural Networks - Jayani Withanawasam
Scene classification using Convolutional Neural Networks - Jayani WithanawasamScene classification using Convolutional Neural Networks - Jayani Withanawasam
Scene classification using Convolutional Neural Networks - Jayani Withanawasam
 
Intelligent image processing
Intelligent image processingIntelligent image processing
Intelligent image processing
 
Anits dip
Anits dipAnits dip
Anits dip
 
Detection and recognition of face using neural network
Detection and recognition of face using neural networkDetection and recognition of face using neural network
Detection and recognition of face using neural network
 
MDEC Data Matters Series: machine learning and Deep Learning, A Primer
MDEC Data Matters Series: machine learning and Deep Learning, A PrimerMDEC Data Matters Series: machine learning and Deep Learning, A Primer
MDEC Data Matters Series: machine learning and Deep Learning, A Primer
 
image_classification.pptx
image_classification.pptximage_classification.pptx
image_classification.pptx
 
slide-171212080528.pptx
slide-171212080528.pptxslide-171212080528.pptx
slide-171212080528.pptx
 
Biometric Recognition using Deep Learning
Biometric Recognition using Deep LearningBiometric Recognition using Deep Learning
Biometric Recognition using Deep Learning
 
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence Spracklen
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence SpracklenSBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence Spracklen
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence Spracklen
 
Real Time Object Dectection using machine learning
Real Time Object Dectection using machine learningReal Time Object Dectection using machine learning
Real Time Object Dectection using machine learning
 
Computer Design Concepts for Machine Learning
Computer Design Concepts for Machine LearningComputer Design Concepts for Machine Learning
Computer Design Concepts for Machine Learning
 
ppt.pdf
ppt.pdfppt.pdf
ppt.pdf
 
Image Segmentation Using Deep Learning : A survey
Image Segmentation Using Deep Learning : A surveyImage Segmentation Using Deep Learning : A survey
Image Segmentation Using Deep Learning : A survey
 
"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...
 
TechnicalBackgroundOverview
TechnicalBackgroundOverviewTechnicalBackgroundOverview
TechnicalBackgroundOverview
 
Deep learning introduction
Deep learning introductionDeep learning introduction
Deep learning introduction
 
Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...
Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...
Deep Learning in Robotics: Robot gains Social Intelligence through Multimodal...
 
Introduction to deep learning
Introduction to deep learningIntroduction to deep learning
Introduction to deep learning
 

Plus de Tulipp. Eu

What are TULIPP starter kits?
What are TULIPP starter kits?What are TULIPP starter kits?
What are TULIPP starter kits?Tulipp. Eu
 
Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...
Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...
Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...Tulipp. Eu
 
HIPPEROS's at EMVA 2017
HIPPEROS's at EMVA 2017 HIPPEROS's at EMVA 2017
HIPPEROS's at EMVA 2017 Tulipp. Eu
 
TULIPP overview
TULIPP overviewTULIPP overview
TULIPP overviewTulipp. Eu
 
HiPEAC 2019 Workshop - Vision Processing
HiPEAC 2019 Workshop - Vision ProcessingHiPEAC 2019 Workshop - Vision Processing
HiPEAC 2019 Workshop - Vision ProcessingTulipp. Eu
 
HiPEAC 2019 Tutorial - Maestro RTOS
HiPEAC 2019 Tutorial - Maestro RTOSHiPEAC 2019 Tutorial - Maestro RTOS
HiPEAC 2019 Tutorial - Maestro RTOSTulipp. Eu
 
HiPEAC 2019 Tutorial - Image Processing Library:HiFlipVX
HiPEAC 2019 Tutorial - Image Processing Library:HiFlipVXHiPEAC 2019 Tutorial - Image Processing Library:HiFlipVX
HiPEAC 2019 Tutorial - Image Processing Library:HiFlipVXTulipp. Eu
 
HiPEAC 2019 Tutorial - Sthem overview
HiPEAC 2019 Tutorial - Sthem overviewHiPEAC 2019 Tutorial - Sthem overview
HiPEAC 2019 Tutorial - Sthem overviewTulipp. Eu
 
HiPEAC 2019 Workshop - Hardware Starter Kit Agri
HiPEAC 2019 Workshop - Hardware Starter Kit Agri HiPEAC 2019 Workshop - Hardware Starter Kit Agri
HiPEAC 2019 Workshop - Hardware Starter Kit Agri Tulipp. Eu
 
HiPEAC 2019 Workshop Overview
HiPEAC 2019 Workshop OverviewHiPEAC 2019 Workshop Overview
HiPEAC 2019 Workshop OverviewTulipp. Eu
 
HiPEAC 2019 Workshop - Use Cases
HiPEAC 2019 Workshop - Use CasesHiPEAC 2019 Workshop - Use Cases
HiPEAC 2019 Workshop - Use CasesTulipp. Eu
 
Tulipp starter-kit-agri
Tulipp starter-kit-agriTulipp starter-kit-agri
Tulipp starter-kit-agriTulipp. Eu
 
TULIPP H2020 Project: Low power high performance real-time computer vision on...
TULIPP H2020 Project: Low power high performance real-time computer vision on...TULIPP H2020 Project: Low power high performance real-time computer vision on...
TULIPP H2020 Project: Low power high performance real-time computer vision on...Tulipp. Eu
 
TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...
TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...
TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...Tulipp. Eu
 
Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017
Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017
Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017Tulipp. Eu
 
D1.1 reference platform_v1_20161215
D1.1 reference platform_v1_20161215D1.1 reference platform_v1_20161215
D1.1 reference platform_v1_20161215Tulipp. Eu
 
Samos July 2016_tulipp-H2020 project presentation
Samos July 2016_tulipp-H2020 project presentationSamos July 2016_tulipp-H2020 project presentation
Samos July 2016_tulipp-H2020 project presentationTulipp. Eu
 
Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016
Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016
Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016 Tulipp. Eu
 

Plus de Tulipp. Eu (18)

What are TULIPP starter kits?
What are TULIPP starter kits?What are TULIPP starter kits?
What are TULIPP starter kits?
 
Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...
Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...
Quantifying Energy Consumption for Practical Fork-Join Parallelism on an Embe...
 
HIPPEROS's at EMVA 2017
HIPPEROS's at EMVA 2017 HIPPEROS's at EMVA 2017
HIPPEROS's at EMVA 2017
 
TULIPP overview
TULIPP overviewTULIPP overview
TULIPP overview
 
HiPEAC 2019 Workshop - Vision Processing
HiPEAC 2019 Workshop - Vision ProcessingHiPEAC 2019 Workshop - Vision Processing
HiPEAC 2019 Workshop - Vision Processing
 
HiPEAC 2019 Tutorial - Maestro RTOS
HiPEAC 2019 Tutorial - Maestro RTOSHiPEAC 2019 Tutorial - Maestro RTOS
HiPEAC 2019 Tutorial - Maestro RTOS
 
HiPEAC 2019 Tutorial - Image Processing Library:HiFlipVX
HiPEAC 2019 Tutorial - Image Processing Library:HiFlipVXHiPEAC 2019 Tutorial - Image Processing Library:HiFlipVX
HiPEAC 2019 Tutorial - Image Processing Library:HiFlipVX
 
HiPEAC 2019 Tutorial - Sthem overview
HiPEAC 2019 Tutorial - Sthem overviewHiPEAC 2019 Tutorial - Sthem overview
HiPEAC 2019 Tutorial - Sthem overview
 
HiPEAC 2019 Workshop - Hardware Starter Kit Agri
HiPEAC 2019 Workshop - Hardware Starter Kit Agri HiPEAC 2019 Workshop - Hardware Starter Kit Agri
HiPEAC 2019 Workshop - Hardware Starter Kit Agri
 
HiPEAC 2019 Workshop Overview
HiPEAC 2019 Workshop OverviewHiPEAC 2019 Workshop Overview
HiPEAC 2019 Workshop Overview
 
HiPEAC 2019 Workshop - Use Cases
HiPEAC 2019 Workshop - Use CasesHiPEAC 2019 Workshop - Use Cases
HiPEAC 2019 Workshop - Use Cases
 
Tulipp starter-kit-agri
Tulipp starter-kit-agriTulipp starter-kit-agri
Tulipp starter-kit-agri
 
TULIPP H2020 Project: Low power high performance real-time computer vision on...
TULIPP H2020 Project: Low power high performance real-time computer vision on...TULIPP H2020 Project: Low power high performance real-time computer vision on...
TULIPP H2020 Project: Low power high performance real-time computer vision on...
 
TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...
TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...
TULIPP H2020 Project presentation @ FPGA Network: Implementing Machine Vision...
 
Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017
Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017
Tulipp_H2020_Hipeac'17 Conference_PEPGUM Workshop_January 017
 
D1.1 reference platform_v1_20161215
D1.1 reference platform_v1_20161215D1.1 reference platform_v1_20161215
D1.1 reference platform_v1_20161215
 
Samos July 2016_tulipp-H2020 project presentation
Samos July 2016_tulipp-H2020 project presentationSamos July 2016_tulipp-H2020 project presentation
Samos July 2016_tulipp-H2020 project presentation
 
Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016
Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016
Tulipp collaboration Workshop - Advanced Computing and CPS - June 2016
 

Dernier

Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...nirzagarg
 
一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理
一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理
一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理uodye
 
Mass storage systems presentation operating systems
Mass storage systems presentation operating systemsMass storage systems presentation operating systems
Mass storage systems presentation operating systemsnight1ng4ale
 
一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证
一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证
一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证wpkuukw
 
一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证
一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证
一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证wpkuukw
 
Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...
Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...
Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...Priya Reddy
 
CRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptx
CRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptxCRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptx
CRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptxRishabh332761
 
怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证
怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证
怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证ehyxf
 
Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...gajnagarg
 
怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证
怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证
怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证tufbav
 
在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一
在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一
在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一ougvy
 
Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...
Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...
Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...vershagrag
 
Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...
Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...
Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...drmarathore
 
Guwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime Guwahati
Guwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime GuwahatiGuwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime Guwahati
Guwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime Guwahatimeghakumariji156
 
🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...
🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...
🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...Call Girls Mumbai
 
一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证
一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证
一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证wpkuukw
 
怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证
怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证
怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证tufbav
 

Dernier (20)

Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Ratlam [ 7014168258 ] Call Me For Genuine Models We...
 
一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理
一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理
一比一维多利亚大学毕业证(victoria毕业证)成绩单学位证如何办理
 
Mass storage systems presentation operating systems
Mass storage systems presentation operating systemsMass storage systems presentation operating systems
Mass storage systems presentation operating systems
 
一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证
一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证
一比一定(购)新西兰林肯大学毕业证(Lincoln毕业证)成绩单学位证
 
一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证
一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证
一比一定(购)UNITEC理工学院毕业证(UNITEC毕业证)成绩单学位证
 
Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...
Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...
Mankhurd Call Girls, 09167354423 Mankhurd Escorts Services,Mankhurd Female Es...
 
CRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptx
CRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptxCRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptx
CRISIS COMMUNICATION presentation=-Rishabh(11195)-group ppt (4).pptx
 
Critical Commentary Social Work Ethics.pptx
Critical Commentary Social Work Ethics.pptxCritical Commentary Social Work Ethics.pptx
Critical Commentary Social Work Ethics.pptx
 
In Riyadh Saudi Arabia |+966572737505 | Buy Cytotec| Get Abortion pills
In Riyadh Saudi Arabia |+966572737505 | Buy Cytotec| Get Abortion pillsIn Riyadh Saudi Arabia |+966572737505 | Buy Cytotec| Get Abortion pills
In Riyadh Saudi Arabia |+966572737505 | Buy Cytotec| Get Abortion pills
 
怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证
怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证
怎样办理阿德莱德大学毕业证(Adelaide毕业证书)成绩单留信认证
 
Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Udgir [ 7014168258 ] Call Me For Genuine Models We ...
 
怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证
怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证
怎样办理斯威本科技大学毕业证(SUT毕业证书)成绩单留信认证
 
在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一
在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一
在线制作(ANU毕业证书)澳大利亚国立大学毕业证成绩单原版一比一
 
Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...
Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...
Low Cost Patna Call Girls Service Just Call 🍑👄6378878445 🍑👄 Top Class Call Gi...
 
Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...
Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...
Abort pregnancy in research centre+966_505195917 abortion pills in Kuwait cyt...
 
Buy Abortion pills in Riyadh |+966572737505 | Get Cytotec
Buy Abortion pills in Riyadh |+966572737505 | Get CytotecBuy Abortion pills in Riyadh |+966572737505 | Get Cytotec
Buy Abortion pills in Riyadh |+966572737505 | Get Cytotec
 
Guwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime Guwahati
Guwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime GuwahatiGuwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime Guwahati
Guwahati Escorts Service Girl ^ 9332606886, WhatsApp Anytime Guwahati
 
🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...
🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...
🌹Patia⬅️ Vip Call Girls Bhubaneswar 📱9777949614 Book Well Trand Call Girls In...
 
一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证
一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证
一比一定(购)坎特伯雷大学毕业证(UC毕业证)成绩单学位证
 
怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证
怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证
怎样办理维多利亚大学毕业证(UVic毕业证书)成绩单留信认证
 

HiPEAC 2019 Workshop - Real-Time Modelling Visual Scenes with Biological Inspiration

  • 1. Real-Time Modelling Visual Scenes with Biological Inspiration Kofi Appiah Sheffield Hallam University
  • 2. AI now and before • Computer Vision and natural language processing have improved significantly over the past 10 years. • Image recognition and classification systems • Apple photo organiser, Facebook face recognition. • Robot use in warehouse • Amazon warehouse robots (https://www.youtube.com/watch?v=4sEVX4mPuto) • Medical image analysis for healthcare • non-invasive diagnosis • Agriculture, sport, manufacturing, autonomous cars technology. • Crop yield, goal-line technology, defective products, people detection. Human level face recognition Taigman et. al. CVPR2014
  • 3. Why AI acceleration • Better algorithms that learn from examples not predefined rules • Deep learning • Neural networks • Machine perception • Availability of data – Big Data • Internet images, YouTube videos, Facebook images • High Performance Computing • Field Programmable Gate Arrays (FPGAs) • Graphics Processing Units (GPU) IEEE Spectrum
  • 4. Key Achievements • Visual recognition with high accuracies. • 3D reconstruction of an environment Mask R-CNN He et. al. ICCV2017 Litjens et. al. 2017 Johnson et. al. CVPR2015 Driverless cars - Mathworks Faster R-CNN TPAMI 2017
  • 5. Where things fall apart • March 18, 2018, Uber’s autonomous car hit and killed 49-year-old as she was walking her bike across the street. • https://www.youtube.com/watch?v=7iTshCm41Ko • Novel and imperfect system • March 23 2018, autopilot Tesla slammed into concrete killing driver. • Security robots attacking a kid in a shopping area, July 2016. • Robot failure to open different doors – which training mode. • Reinforced learning • Supervised or Unsupervised?
  • 6. Why things go wrong • For autonomous cars, the state of the art is good and providing bounding boxes of objects in the scene. • What is missing is an interpretation of the scene. • No contextual reasoning. • Robot navigation • Decision making might be optimal but not feasible or safe. • Modelling in a crowded scene to infer interaction • Modelling very unusual situations with little or no data • Things that human are capable of, e.g. dealing with complex scenes Fei-Fei Li
  • 7. Unsupervised Background Subtraction • Image Segmentation separate moving objects from the background. • Background subtraction is a practical approach when the image sensor is stationary. • Background Modelling techniques - Unimodal - Multimodal
  • 8. W4 and Grimson’s Algorithm – 2000s • Requires manual initialization of the Maximum (M), Minimum (m) & inter-frame difference (D) • Pixel x of image I is foreground if |m(x)-It(x)|>D(x) or |M(x)- It(x)|>D(x) • Detection, Motion & change history maps used for outdoor scene. • Use of fixed-point update values. • Bimodal can’t model problems like moving foliage and lighting changes. • Mixture of Gaussians with associated weights to model each pixel. • Parameters are updated as follows: • The first B distributions, ordered by weight represents the background • Robust in modeling multimodal background. • Suffers from blending effect and uses floating point in all updates
  • 9. Efficient Hardware Implementation • Maintains K clusters each with weight wk, central value ck and implied global range [ck-15, ck+15] • Weights and central values of all clusters are initialized 0, and updated as follows: • Uses both pixel and frame-level processing • The first B distributions, ordered by weight represents the background       + = − − otherwise 64 63 clustermatchingfor the 64 1 64 63 1, 1, , tk tk tk         + = − − otherwise clustermatching 8 1 8 7 ,,1, ,,,1, ,,, jitk jijitk jitk c Xc c       = = b k ib TB 1 minarg  Appiah et al FPT 2005
  • 10. TULIPP – The game changer! • Tools to help real-time computer vision developer to focus on: • core application development by automating recurring, but critical, tasks such as performance instrumentation • Design space exploration and • Vendor tool configuration. • Making it possible for the designer to get the required performance in speed, coupled with power constraints without having to worry too much about the architecture.
  • 11. Imaging before Deep Learning Before • Standard feature detectors • SIFT, HOG, LBP • Different algorithms for object detection • Requires small amount of data • Useful for measurement and labelling After • Featured are learnt and stacked according to data • Same algorithm that adapts to the data • Requires huge volume of data • Useful for labelling MathworksDalal & Triggs cc.gatech.edu
  • 12. Deep CNN – Overview • Uses convolution to preserve the spatial structure of the input image • Instead of a sigmoid activation function, ReLU (rectified linear unit) is often used • Encourages sparsity of synapses as the value approaches zero (0). Credit : Fei-Fei Li CS231n; Bala Amavasai – IEEE & M. Turner
  • 13. Feature Maps - Several feature maps are used to identify various local features • Several feature maps are used to identify various local features. • Each convolution filter can be tuned to edges of different • Orientation, Frequency, Phase, Colour, etc • Capture some aspects of neural response • But neural data not used in training
  • 14. Sparse local connectivity • For an input image of size 7x7 • The convolution filter 3x3 • The output image will be 5x5 • (Image – Filter )/stride + 1 • A sample filter for horizontal and vertical gradient.
  • 15. Way forward • Computer Vision meets Cognitive Science and Neuroscience Fei-Fei Li & Justin Johnson & Serena Yeung
  • 16. The Challenge • The success stories about the rise of Convolutional Neural Networks (CNNs) capable of learning high-level features in object recognition keeps increasing • due to the availability of large datasets like ImageNet • However, performance at scene recognition has not attained the same level of success. • Yet large scene databases like SUN and Places do exist • Maybe the current deep features trained from ImageNet are not competitive enough for such tasks. • But do primates and humans actually do a raster scan to understand a scene? • CNNs fail to capture insensitivity to perturbations of an image
  • 17. Possible Solution • Performance accuracies in CNNs relies on a huge search space. • The need for more biological guidance from the visual cortex • Multi-disciplinary research in neuroscience, psychology, physiology, shows that: • object recognition in visual cortex is modulated via the ventral stream • Neuronal signals from the retina are transformed into high-level representation for object recognition. • Computer Scientist working with neuroscientist, psychologist, etc. would have better models for understanding scenes.
  • 18. Reported Successes • A biologically Inspired Deep CNN Model [Zhang et al. 2016] • Simulates the V1, V2, V4 and IT layers of the human ventral stream • Uses convolutional layers with varied sizes and complexities • Increased concurrency for improved processing speed • Outperformed seven other CNN techniques using four datasets. • You Only Look Once (YOLOv2) [Redmon and Farhadi CVPR2017] • Based on the assumption that humans glance at an image • Does not rely on sliding window like other deep learning approaches • Outperforms Deformable Part Models (DPM) and Regional CNN.
  • 19. Scene understanding with DNN • Learning Deep Features for Scene Recognition using Places Database [Zhou et al. NIPS2014] • Uses CNN to learn features from the scene • Combined various local and global features to understand the scene • Presents scene categories where machines perform like humans. • Humans, but Not Deep Neural Networks, Often Miss Giant Targets in Scenes [Eckstein et al. Current Biology 2017] • Humans often miss unusual sized targets during visual search • Deep learning does not exhibit such deficit with targets • Is that a good thing or not?
  • 20. Our motivation • Missing giant targets is a functional brain strategy to discount distractors Eckstein et al. Current Biology 2017
  • 21. Our Approach • To understand how humans and primates recognise scenes • Provide them with samples of indoor scenes • Ask them to identify specific objects • Observe their recall mechanism, if spatial relationship plays a role • Model the scene to account for the experimental results • Incorporate global and local descriptors • Construct a relationship vector Lunchroom image : PASSTA Dataset
  • 22. Summary • Computer vision and machine learning have improved over the years, thanks to more data and processing power. • Global scene understanding is still a challenge. • Multi-disciplinary effort required to take computer vision to the next level, acceptable for applications like driverless cars. • We aim to combine positives of CNN with what humans are good at for scene understanding. • TULIPP offers the platform with toolchain to drive this agenda.