# Open-World Object Detection via Discriminative Class Prototype Learning

## TL;DR

This paper introduces OCPL, a novel approach for open-world object detection that learns discriminative class prototypes to effectively identify known and unknown objects, improving detection and incremental learning.

## Contribution

The paper proposes a prototype-based framework with modules like PEA, ESC, and CSC to enhance discriminative embeddings for open-world object detection.

## Key findings

- Effective differentiation of known and unknown classes.
- Improved detection performance on PASCAL VOC and MS-COCO.
- Enhanced incremental learning capabilities.

## Abstract

Open-world object detection (OWOD) is a challenging problem that combines object detection with incremental learning and open-set learning. Compared to standard object detection, the OWOD setting is task to: 1) detect objects seen during training while identifying unseen classes, and 2) incrementally learn the knowledge of the identified unknown objects when the corresponding annotations is available. We propose a novel and efficient OWOD solution from a prototype perspective, which we call OCPL: Open-world object detection via discriminative Class Prototype Learning, which consists of a Proposal Embedding Aggregator (PEA), an Embedding Space Compressor (ESC) and a Cosine Similarity-based Classifier (CSC). All our proposed modules aim to learn the discriminative embeddings of known classes in the feature space to minimize the overlapping distributions of known and unknown classes, which is beneficial to differentiate known and unknown classes. Extensive experiments performed on PASCAL VOC and MS-COCO benchmark demonstrate the effectiveness of our proposed method.