YOLOX-S detector model and weights
Copyright (c) Megvii, Inc.
Licensed under Apache License 2.0.
Source revision: 6ddff4824372906469a7fae2dc3206c7aa4bbaee
Source: https://github.com/Megvii-BaseDetection/YOLOX

Qualcomm AI Hub Models export wrapper and model metadata
Copyright (c) Qualcomm Innovation Center, Inc.
Licensed under the BSD 3-Clause License.
Release: v0.61.0, commit 5975a79b55b40f5cbc61f3ac5e52abe47d9d8bd5
Source: https://github.com/qualcomm/ai-hub-models

Selected asset:
https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolox/releases/v0.61.0/yolox-onnx-float.zip
SHA-256: 3f0c52589814aeddf7df888c91f9bd0f0025fa80f93012ce3769901f129485eb
Qualcomm model metadata identifies the model license as Apache-2.0 and the
training dataset as COCO.

COCO annotations and website content are licensed under CC BY 4.0. COCO does
not own the training images; those remain governed by their Flickr image
licenses. No COCO training image is redistributed in this package.

The 16 calibration frames are COCO test2017 entries whose official image-info
metadata marks license id 7, Flickr Commons "No known copyright restrictions."
Their exact URLs and SHA-256 identities are retained in the packaged preparation
report. Calibration image bytes are not redistributed in this package.

The Qualcomm AI Runtime / QNN and ONNX Runtime binaries are not covered by the
model license. Their local use and any public/customer redistribution remain
subject to their own notices and ModalAI legal/export approval.
