Robust category-level 6d pose estimation with coarse-to-fine rendering of neural features
We consider the problem of category-level 6D pose estimation from a single RGB image.
Our approach represents an object category as a cuboid mesh and learns a generative …
Our approach represents an object category as a cuboid mesh and learns a generative …
How Shape Perception Works, in Two Dimensions and Three Dimensions
KJ Nielsen, CE Connor - Annual Review of Vision Science, 2024 - annualreviews.org
The ventral visual pathway transforms retinal images into neural representations that
support object understanding, including exquisite appreciation of precise 2D pattern shape …
support object understanding, including exquisite appreciation of precise 2D pattern shape …
Neural view synthesis and matching for semi-supervised few-shot learning of 3d pose
We study the problem of learning to estimate the 3D object pose from a few labelled
examples and a collection of unlabelled data. Our main contribution is a learning framework …
examples and a collection of unlabelled data. Our main contribution is a learning framework …
Unsupervised causal generative understanding of images
T Anciukevicius, P Fox-Roberts… - Advances in …, 2022 - proceedings.neurips.cc
We present a novel framework for unsupervised object-centric 3D scene understanding that
generalizes robustly to out-of-distribution images. To achieve this, we design a causal …
generalizes robustly to out-of-distribution images. To achieve this, we design a causal …
BIFRNet: a brain-inspired feature restoration DNN for partially occluded image recognition
The partially occluded image recognition (POIR) problem has been a challenge for artificial
intelligence for a long time. A common strategy to handle the POIR problem is using the non …
intelligence for a long time. A common strategy to handle the POIR problem is using the non …
[PDF][PDF] Towards Bayesian Generative Architectures for Out-of-Distribution Tasks
A Yuille - neural-architecture-ppf.github.io
Why We Need better Architectures Page 1 Towards Bayesian Generative Architectures for
Out-ofDistribution Tasks Alan Yuille Bloomberg Distinguished Professor Depts. Computer …
Out-ofDistribution Tasks Alan Yuille Bloomberg Distinguished Professor Depts. Computer …
[CITATION][C] Repetition sans Ennui: Human and Algorithmic Discovery of Patterns in Music
IY Ren - 2024 - Utrecht University