Scaling Novel Object Detection with Weakly Supervised Detection Transformers

TL;DR

This paper introduces a Weakly Supervised Detection Transformer that enhances novel object detection by leveraging large-scale pretraining, improving efficiency and performance over existing WSOD methods, especially in large-scale scenarios.

Contribution

The paper presents a novel WSOD framework using transformers that enables effective knowledge transfer from large pretraining datasets to detect many novel objects.

Findings

Outperforms previous state-of-the-art WSOD models on large-scale datasets

Class quantity is more crucial than image quantity for WSOD pretraining

The proposed method reduces training rounds and refinement steps

Abstract

A critical object detection task is finetuning an existing model to detect novel objects, but the standard workflow requires bounding box annotations which are time-consuming and expensive to collect. Weakly supervised object detection (WSOD) offers an appealing alternative, where object detectors can be trained using image-level labels. However, the practical application of current WSOD models is limited, as they only operate at small data scales and require multiple rounds of training and refinement. To address this, we propose the Weakly Supervised Detection Transformer, which enables efficient knowledge transfer from a large-scale pretraining dataset to WSOD finetuning on hundreds of novel objects. Additionally, we leverage pretrained knowledge to improve the multiple instance learning (MIL) framework often used in WSOD methods. Our experiments show that our approach outperforms previous state-of-the-art models on large-scale novel object detection datasets, and our scaling study reveals that class quantity is more important than image quantity for WSOD pretraining. The code is available at https://github.com/tmlabonte/weakly-supervised-DETR.