DetCLIPv3: Towards Versatile Generative Open-vocabulary Object Detection

TL;DR

DetCLIPv3 is a versatile open-vocabulary object detector that also generates hierarchical labels, leveraging a robust architecture, rich training data, and efficient strategies to outperform existing models in detection and captioning tasks.

Contribution

The paper introduces DetCLIPv3, a novel model combining open-vocabulary detection with hierarchical caption generation, enhanced by auto-annotation and a two-stage training process.

Findings

Achieves 47.0 zero-shot AP on LVIS benchmark.

Outperforms previous models like GLIPv2 and GroundingDINO in detection.

Sets new state-of-the-art in dense captioning with 19.7 AP on VG dataset.

Abstract

Existing open-vocabulary object detectors typically require a predefined set of categories from users, significantly confining their application scenarios. In this paper, we introduce DetCLIPv3, a high-performing detector that excels not only at both open-vocabulary object detection, but also generating hierarchical labels for detected objects. DetCLIPv3 is characterized by three core designs: 1. Versatile model architecture: we derive a robust open-set detection framework which is further empowered with generation ability via the integration of a caption head. 2. High information density data: we develop an auto-annotation pipeline leveraging visual large language model to refine captions for large-scale image-text pairs, providing rich, multi-granular object labels to enhance the training. 3. Efficient training strategy: we employ a pre-training stage with low-resolution inputs that enables the object captioner to efficiently learn a broad spectrum of visual concepts from extensive image-text paired data. This is followed by a fine-tuning stage that leverages a small number of high-resolution samples to further enhance detection performance. With these effective designs, DetCLIPv3 demonstrates superior open-vocabulary detection performance, \eg, our Swin-T backbone model achieves a notable 47.0 zero-shot fixed AP on the LVIS minival benchmark, outperforming GLIPv2, GroundingDINO, and DetCLIPv2 by 18.0/19.6/6.6 AP, respectively. DetCLIPv3 also achieves a state-of-the-art 19.7 AP in dense captioning task on VG dataset, showcasing its strong generative capability.