UC-OWOD: Unknown-Classified Open World Object Detection (2022-07-23T00:00:00.000000Z)

TL;DR

This work formulate the problem and devise a two-stage object detector to solve UC-OWOD, a novel OWOD problem called Unknown-Classified Open World Object Detection, which aims to detect unknown instances and classify them into different unknown classes.

Abstract

Open World Object Detection (OWOD) is a challenging computer vision problem that requires detecting unknown objects and gradually learning the identified unknown classes. However, it cannot distinguish unknown instances as multiple unknown classes. In this work, we propose a novel OWOD problem called Unknown-Classified Open World Object Detection (UC-OWOD). UC-OWOD aims to detect unknown instances and classify them into different unknown classes. Besides, we formulate the problem and devise a two-stage object detector to solve UC-OWOD. First, unknown label-aware proposal and unknown-discriminative classification head are used to detect known and unknown objects. Then, similarity-based unknown classification and unknown clustering refinement modules are constructed to distinguish multiple unknown classes. Moreover, two novel evaluation protocols are designed to evaluate unknown-class detection. Abundant experiments and visualizations prove the effectiveness of the proposed method. Code is available at https://github.com/JohnWuzh/UC-OWOD.

Authors

Zhi-zong Wu

1 papers

Yue Lu

1 papers

Xingyu Chen

1 papers

TL;DR

Abstract

Authors

References62 items

Decoupled Metric Network for Single-Stage Few-Shot Object Detection

Revisiting Open World Object Detection

OW-DETR: Open-world Detection Transformer

Bridging Non Co-occurrence with Unlabeled In-the-wild Data for Incremental Object Detection

J-score: a robust measure of clustering accuracy

Detection and Continual Learning of Novel Face Presentation Attacks

Wanderlust: Online Continual Object Detection in the Real World

Towards Open World Object Detection

Counterfactual Zero-Shot and Open-Set Visual Recognition

Deformable DETR: Deformable Transformers for End-to-End Object Detection

Meta-Consolidation for Continual Learning

GDumb: A Simple Approach that Questions Our Progress in Continual Learning

D2Det: Towards High Quality Object Detection and Instance Segmentation

Generative-Discriminative Feature Representations for Open-Set Recognition

iTAML: An Incremental Task-Agnostic Meta-learning Approach

Incremental Object Detection via Meta-Learning

Frustratingly Simple Few-Shot Object Detection

Incremental Few-Shot Object Detection

The Overlooked Elephant of Object Detection: Open Set

Scalability in Perception for Autonomous Driving: Waymo Open Dataset

Discovering New Intents via Constrained Deep Adaptive Clustering with Cluster Refinement

Large-Scale Long-Tailed Recognition in an Open World

nuScenes: A Multimodal Dataset for Autonomous Driving

Classification-Reconstruction Learning for Open-Set Recognition

Probabilistic Object Detection: Definition and Evaluation

Evaluating Merging Strategies for Sampling-based Uncertainty Techniques in Object Detection

Open-world Learning and Application to Product Classification

End-to-End Incremental Learning

Joint Anchor-Feature Refinement for Real-Time Accurate Object Detection in Images and Videos

Zero-Shot Object Detection

Overcoming catastrophic forgetting with hard attention to the task

Learning to cluster in order to Transfer across domains and tasks

Memory Aware Synapses: Learning what (not) to forget

PackNet: Adding Multiple Tasks to a Single Network by Iterative Pruning

Dropout Sampling for Robust Object Detection in Open-Set Conditions

Focal Loss for Dense Object Detection

Mask R-CNN

iCaRL: Incremental Classifier and Representation Learning

Visual Relationship Detection with Language Priors

Learning without Forgetting

Progressive Neural Networks

Deep Residual Learning for Image Recognition

Unsupervised Deep Embedding for Clustering Analysis

Towards Open Set Deep Networks

You Only Look Once: Unified, Real-Time Object Detection

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Fast R-CNN

Towards Open World Recognition

Multi-class Open Set Recognition Using Probability of Inclusion

Probability Models for Open Set Recognition

Microsoft COCO: Common Objects in Context

Vision meets robotics: The KITTI dataset

Toward Open Set Recognition

Constrained K-means with external information

Deep learning for detecting robotic grasps

Active Semi-Supervision for Pairwise Constrained Clustering

Catastrophic forgetting in connectionist networks

The Hungarian method for the assignment problem

A Survey of Constrained Clustering

Visualizing Data using t-SNE

Catastrophic Interference in Connectionist Networks: The Sequential Learning Problem

Edinburgh Research Explorer The PASCAL Visual Object Classes (VOC) Challenge

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names