Overview
Explore the fascinating world of visual perception and material properties in this 41-minute lecture by Dr. Vivian Paulun from MIT's Kanwisher and Tenenbaum Labs. Delve into the complexities of how the human brain infers properties like mass, compliance, elasticity, and fragility from visual cues. Examine experimental results on human observers' ability to estimate elasticity and predict the future path of bouncing objects. Analyze the visual information used by observers and the limitations of their internal models. Investigate motion features, principal component analysis, and the role of perception in understanding material behavior. Compare prediction and reconstruction processes, and learn about the effects of training on accuracy and consistency. Gain insights into the simulation of material properties and their impact on our understanding of the physical world.
Syllabus
Intro
Visual estimation of material properties
Elastic bouncing cubes
Experimental results
Motion features
Principal component analysis
Perception
Results
Summary
Prediction vs Reconstruction
Accuracy with Training
Consistency
Simulation
Conclusion
Taught by
MITCBMM