Simple Open-Vocabulary Object Detection with Vision Transformers
This addresses the problem of object detection with limited training data for researchers and practitioners, though it is incremental as it builds on existing architectures and pre-training methods.
The paper tackles open-vocabulary object detection by proposing a recipe that transfers image-text models to this task, achieving strong performance in zero-shot text-conditioned and one-shot image-conditioned settings.
Combining simple architectures with large-scale pre-training has led to massive improvements in image classification. For object detection, pre-training and scaling approaches are less well established, especially in the long-tailed and open-vocabulary setting, where training data is relatively scarce. In this paper, we propose a strong recipe for transferring image-text models to open-vocabulary object detection. We use a standard Vision Transformer architecture with minimal modifications, contrastive image-text pre-training, and end-to-end detection fine-tuning. Our analysis of the scaling properties of this setup shows that increasing image-level pre-training and model size yield consistent improvements on the downstream detection task. We provide the adaptation strategies and regularizations needed to attain very strong performance on zero-shot text-conditioned and one-shot image-conditioned object detection. Code and models are available on GitHub.