Multi-Modal Classifiers for Open-Vocabulary Object Detection