Learning to See by Moving
E326789
"Learning to See by Moving" is a research work in computer vision that explores how visual understanding can emerge from an agent’s own movement and interaction with the environment, rather than from static images alone.
All labels observed (1)
| Label | Occurrences |
|---|---|
| Learning to See by Moving canonical | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T3094205 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: Learning to See by Moving Context triple: [Alexei Efros, notableWork, Learning to See by Moving]
-
A.
CMU Highly Intelligent Mobile Platform
CMU Highly Intelligent Mobile Platform (CHIMP) is a sophisticated humanoid robot developed at Carnegie Mellon University for advanced mobility, manipulation, and autonomous operation in challenging environments.
-
B.
Technical Committee on Robot Learning
The Technical Committee on Robot Learning is a specialized IEEE Robotics and Automation Society group that advances research and collaboration at the intersection of machine learning and robotics.
-
C.
Robot Visions
Robot Visions is a collection of Isaac Asimov’s robot-themed short stories and essays that explores his famous Three Laws of Robotics and their implications.
-
D.
Soundtrack to Human Motion
Soundtrack to Human Motion is the acclaimed debut jazz album by American pianist and composer Jason Moran, noted for its innovative blend of tradition and modern experimentation.
-
E.
SLAM
SLAM is a major art museum in St. Louis, Missouri, renowned for its extensive collection spanning thousands of years and diverse cultures.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: Learning to See by Moving Target entity description: "Learning to See by Moving" is a research work in computer vision that explores how visual understanding can emerge from an agent’s own movement and interaction with the environment, rather than from static images alone.
-
A.
CMU Highly Intelligent Mobile Platform
CMU Highly Intelligent Mobile Platform (CHIMP) is a sophisticated humanoid robot developed at Carnegie Mellon University for advanced mobility, manipulation, and autonomous operation in challenging environments.
-
B.
Technical Committee on Robot Learning
The Technical Committee on Robot Learning is a specialized IEEE Robotics and Automation Society group that advances research and collaboration at the intersection of machine learning and robotics.
-
C.
Robot Visions
Robot Visions is a collection of Isaac Asimov’s robot-themed short stories and essays that explores his famous Three Laws of Robotics and their implications.
-
D.
Soundtrack to Human Motion
Soundtrack to Human Motion is the acclaimed debut jazz album by American pianist and composer Jason Moran, noted for its innovative blend of tradition and modern experimentation.
-
E.
SLAM
SLAM is a major art museum in St. Louis, Missouri, renowned for its extensive collection spanning thousands of years and diverse cultures.
- F. None of above. chosen
Statements (40)
| Predicate | Object |
|---|---|
| instanceOf |
computer vision research work
ⓘ
research paper ⓘ |
| aimsTo |
learn depth cues from movement
ⓘ
learn object boundaries from motion parallax ⓘ learn predictive visual models ⓘ learn scene structure from motion ⓘ |
| arguesThat | visual understanding can emerge from interaction with the environment ⓘ |
| assumes |
an embodied agent with control over its motion
ⓘ
continuous interaction with the environment ⓘ |
| challenges | purely static dataset-based training paradigms ⓘ |
| comparesWith | representations learned from static images ⓘ |
| contrastsWith | learning from static images alone ⓘ |
| contributesTo |
methods for learning world models from interaction
ⓘ
understanding of how agents can autonomously acquire visual skills ⓘ |
| demonstrates |
that agents can improve perception by exploring
ⓘ
that motion provides supervisory signals for vision ⓘ |
| emphasizes |
the coupling between perception and action
ⓘ
the role of temporal continuity in vision ⓘ |
| evaluates | visual representations learned from motion ⓘ |
| field |
computer vision
ⓘ
machine learning ⓘ robotics ⓘ |
| focusesOn |
active perception
ⓘ
embodied perception ⓘ self-supervised learning from motion ⓘ sensorimotor learning ⓘ visual understanding ⓘ |
| inspiredBy | how animals learn vision through movement ⓘ |
| proposes | learning visual representations from an agent’s own movement ⓘ |
| relatedTo |
active vision
ⓘ
developmental robotics ⓘ embodied AI ⓘ reinforcement learning for perception ⓘ self-supervised representation learning ⓘ |
| shows |
that interaction can provide intrinsic supervision for vision
ⓘ
that motion-based learning can capture 3D structure ⓘ |
| uses |
an agent that moves in an environment
ⓘ
egocentric visual observations ⓘ sequences of images over time ⓘ the agent’s own actions as supervision signal ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: Learning to See by Moving Description of subject: "Learning to See by Moving" is a research work in computer vision that explores how visual understanding can emerge from an agent’s own movement and interaction with the environment, rather than from static images alone.
Referenced by (1)
Full triples — surface form annotated when it differs from this entity's canonical label.