Description Logic for VisionBased - PowerPoint PPT Presentation

1 / 31
About This Presentation
Title:

Description Logic for VisionBased

Description:

Learn. Generic. Geometry. Model. Logical 'Configuration' Model. DL Road. Network KB ... KB will not be fully unambiguous due to required 'hacks' 30. Outlook ... – PowerPoint PPT presentation

Number of Views:30
Avg rating:3.0/5.0
Slides: 32
Provided by: dags3
Category:

less

Transcript and Presenter's Notes

Title: Description Logic for VisionBased


1
  • Description Logic for Vision-Based
  • Intersection Interpretation

Britta Hummel
2
Motivation
Road Recognition The Model-based Approach
1. Project
Low-dim. geometry model (clothoid, )
2. Compare
3. Update Parameters
  • Solved for highly constrained domains (highways)

3
Motivation
Intersection Recognition HeimesNagel02
1. Project
2. Compare
3. Update Parameters
  • How can we generalize to arbitrary intersections?

4
Motivation
Challenges
  • High-dimensional
  • hypothesis space
  • 2. Few features
  • - Narrow field of view
  • - Massive occlusions
  • - Omitted features
  • Presence of noise
  • - Unmodelled objects
  • - Bad feature quality
  • Model-based approach becomes ill-posed!

5
Motivation
So what now?
  • More top-down information flow
  • ? start higher up use conceptual knowledge!
  • ? move further down parameterize feature
    detectors!
  • Collective classification
  • ? simultaneously reconstruct geometry and
    semantics!
  • Narrow down hypothesis space!
  • FOL Representation and FOL Reasoning!

6
This Talk
  • Motivation
  • Architecture
  • DL Road Network KB
  • DL Inference for Scene Interpretation
  • Application
  • Evaluation

7
Architecture
Enhance Model-Based Vision by Logic
Feature detectors, other KBs,
Project
Update Pars
8
This Talk
  • Motivation
  • Architecture
  • DL Road Network KB
  • DL Inference for Scene Interpretation
  • Application
  • Evaluation

9
Model of Geometry
Geometric Primitives
GP1
GP3
GP2
Spatial Relations
10
Symbol Grounding
Geometric Primitives
Spatial Relations
11
TBox
Geometric Constraints
12
TBox
Constraints wrt Road Building Regulations
13
ABox
Sensor Data Integration
  • Partial observability
  • ? OWA
  • Structurally differing sensor data (e.g. from
    map, video)
  • Distributed sensor data
  • Non-UNA identification reasoning
  • Open/Closed Domain Data
  • (Nominals) / Closed domain assumption
  • Conflicting/Uncertain Data
  • ? BLPs/MLNs/

14
This Talk
  • Motivation
  • Architecture
  • DL Road Network KB
  • DL Inference for Scene Interpretation
  • Application
  • Evaluation

15
Inference Example I
(Collective) Classification is Abox realization
l21
l22
l11
l12
l13
16
Inference Example I
(Collective) Classification is Abox realization
l21
l22
l11
l12
l13
16
17
Inference Example I
(Collective) Classification is Abox realization
l21
l22
tr-l11-l21
l11
l12
l13
17
18
Inference Example I
(Collective) Classification is Abox realization
l21
l22
tr-l11-l21

l11
l12
l13
19
Inference Example I
Link Prediction is Instance Checking
19
20
Inference Example II
Link Prediction is Instance Checking
l21
l22
tr-l11-l21
l11
l12
l13
21
Inference Example III
Data Association is Identification Reasoning
Positioning Device Map Matching
Video
Digital Map
?
22
Inference Example IV
Hypothesis Generation is ?
  • Classical logical inference is deductive
  • Bio./Mach. Vision is not deductive lots of
    hypothetical reasoning, jumping to conclusions,
    backtracking if wrong
  • ? Non-deductive / non-monotonic reasoning
    needed!
  • Abduction
  • Poole, Shanahan, Möller
  • Introducing procedurality
  • NeumannMöller06
  • Model construction by transformation into
    Constraint Satisfaction Pr.
  • ReiterMackworth87
  • Model construction under Answer Set Semantics
  • We have started

23
Beautiful Analogies
24
This Talk
  • Motivation
  • Architecture
  • DL Road Network KB
  • DL Inference for Scene Interpretation
  • Application
  • Evaluation

25
Application
Geometry model generated from DL ground truth
ABox
26
Application
27
This Talk
  • Motivation
  • Architecture
  • DL Road Network KB
  • DL Inference for Scene Interpretation
  • Application
  • Evaluation

28
Summary
  • Road Recognition ? Intersection Interpretation
  • escape from toy world ? narrow down hypothesis
    space
  • not only bottom-up but also top-down
    reasoning
  • collective classification
  • Enhance model-based vision by logical reasoning
  • Expressive geometry model
  • Generate generic geometric model out of logical
    configuration model
  • Generate and constrain logical model through
    logical reasoning

29
Evaluation
  • Vision
  • Sets of knowledge engineers codingmaintaining
    large, distributed, modular, semantically
    unambiguous KBs for SI
  • DL
  • Wish List ?
  • Foundational ontologies / best practices for KB
    design for SI
  • Faster Abox reasoning (gt10 individuals
    prohibitively slow on our KB)
  • Language expressiveness
  • Spatial Relations JEPD condition
  • Feature chains
  • Nominals
  • Nonmonotonic reasoning

30
Outlook
  • Nonmonotonic reasoning with ASP
  • Incremental hypothesize test
  • Integration with Irina Lulchevas MLN-based
    traffic participant classificator
  • Rule Learning from Training Data

31
Thanks
  • Thanks ?

32
Description Logic
  • Decidable subset of 1st order logic
  • Syntax
  • Semantics Set-theoretic

33
Description Logic
  • Axioms form sentences
  • A DL Knowledge Base consists of
  • Tbox Set of terminological axioms
  • ? general domain knowledge
  • Abox Set of assertional axioms
  • ? knowledge about particular situation
  • ( Rulebox )

34
Description Logic
  • Classical DL inference services
  • Non-classical inference
  • .
Write a Comment
User Comments (0)
About PowerShow.com