OWLv2 -- large
open-vocabulary object detection, trained by self-training on Web image-text pairs (ST)

Copyright 2023 The Google Research Authors.
Matthias Minderer, Alexey Gritsenko, Neil Houlsby.

Source:  https://huggingface.co/google/owlv2-large-patch14/resolve/main/pytorch_model.bin
Project: https://github.com/google-research/scenic/tree/main/scenic/projects/owl_vit
Paper:   Scaling Open-Vocabulary Object Detection (NeurIPS 2023), arXiv:2306.09683
Licence: Apache-2.0 (full text in the LICENSE file beside this one)

Google's own release is JAX/Flax. This file is the PyTorch conversion Google publishes
beside it on Hugging Face, unmodified, under the same terms.