Twenty Thousand Class

"Twenty-thousand class" research focuses on expanding the capabilities of computer vision models to recognize and segment a vastly increased number of object categories, significantly exceeding the limitations of traditional datasets. Current efforts involve integrating powerful segmentation models like SAM with zero-shot recognition models like CLIP, and developing novel algorithms like Invariant-Feature Subspace Recovery (ISR) to improve generalization across diverse visual domains. This research is crucial for advancing open-vocabulary visual recognition, enabling applications such as more robust object detection, improved image understanding, and more accurate scene interpretation in various fields.

Papers

March 4, 2024

Exponential Expressivity of ReLU$^k$ Neural Networks on Gevrey Classes with Point Singularities
Joost A. A. Opschoor, Christoph Schwab
Neural Network Singularity Free Expressivity Transfer Piecewise Polynomial Polygonal Domain Twenty Thousand Class

January 5, 2024

Open-Vocabulary SAM: Segment and Recognize Twenty-thousand Classes Interactively
Haobo Yuan, Xiangtai Li, Chong Zhou, Yining Li, Kai Chen, Chen Change Loy
Zero Shot Segment Anything Model Segmentation Task Well Defined Segment Twenty Thousand Class

November 2, 2023

Invariant-Feature Subspace Recovery: A New Class of Provable Domain Generalization Algorithms
Haoxiang Wang, Gargi Balasubramaniam, Haozhe Si, Bo Li, Han Zhao
Domain Generalization Invariant Risk Minimization Invariant Subspace Twenty Thousand Class

May 22, 2023

Squared Neural Families: A New Class of Tractable Density Models
Russell Tsuchida, Cheng Soon Ong, Dino Sejdinovic
Density Estimation Neural Process Conditional Density Tractable Density Twenty Thousand Class Tractable Initial Distribution

April 28, 2023

A New Class of Explanations for Classifiers with Non-Binary Features
Chunxi Ji, Adnan Darwiche
Line by Line Explanation Counterfactual Explanation Simple Classifier Abductive Explanation Twenty Thousand Class

April 10, 2023

Prompt Pre-Training with Twenty-Thousand Classes for Open-Vocabulary Visual Recognition
Shuhuai Ren, Aston Zhang, Yi Zhu, Shuai Zhang, Shuai Zheng, Mu Li, Alex Smola, Xu Sun
Semantic Segmentation Vision Language Model Recognition Task Twenty Thousand Class

January 7, 2022

Detecting Twenty-thousand Classes using Image-level Supervision
Xingyi Zhou, Rohit Girdhar, Armand Joulin, Philipp Krähenbühl, Ishan Misra
Object Detector ImageNet Dataset Image Classifier Level Supervision Image Classification Datasets Detection Datasets Twenty Thousand Class

Twenty Thousand Class

Papers

Exponential Expressivity of ReLU$^k$ Neural Networks on Gevrey Classes with Point Singularities

Open-Vocabulary SAM: Segment and Recognize Twenty-thousand Classes Interactively

Invariant-Feature Subspace Recovery: A New Class of Provable Domain Generalization Algorithms

Squared Neural Families: A New Class of Tractable Density Models

A New Class of Explanations for Classifiers with Non-Binary Features

Prompt Pre-Training with Twenty-Thousand Classes for Open-Vocabulary Visual Recognition

Detecting Twenty-thousand Classes using Image-level Supervision