add AIBOM

Dear model owner(s),
We are a group of researchers investigating the usefulness of sharing AIBOMs (Artificial Intelligence Bill of Materials) to document AI models – AIBOMs are machine-readable structured lists of components (e.g., datasets and models) used to enhance transparency in AI-model supply chains.

To pursue the above-mentioned objective, we identified popular models on HuggingFace and, based on your model card (and some configuration information available in HuggingFace), we generated your AIBOM according to the CyclonDX (v1.6) standard (see https://cyclonedx.org/docs/1.6/json/). AIBOMs are generated as JSON files by using the following open-source supporting tool: https://github.com/MSR4SBOM/ALOHA (technical details are available in the research paper: https://github.com/MSR4SBOM/ALOHA/blob/main/ALOHA.pdf).

The JSON file in this pull request is your AIBOM (see https://github.com/MSR4SBOM/ALOHA/blob/main/documentation.json for details on its structure).

Clearly, the submitted AIBOM matches the current model information, yet it can be easily regenerated when the model evolves, using the aforementioned AIBOM generator tool.

We open this pull request containing an AIBOM of your AI model, and hope it will be considered. We would also like to hear your opinion on the usefulness (or not) of AIBOM by answering a 3-minute anonymous survey: https://forms.gle/WGffSQD5dLoWttEe7.

Thanks in advance, and regards,
Riccardo D’Avino, Fatima Ahmed, Sabato Nocera, Simone Romano, Giuseppe Scanniello (University of Salerno, Italy),
Massimiliano Di Penta (University of Sannio, Italy),
The MSR4SBOM team

google_owlv2-base-patch16-ensemble.json

@@ -0,0 +1,61 @@
+{
+    "bomFormat": "CycloneDX",
+    "specVersion": "1.6",
+    "serialNumber": "urn:uuid:e5046994-9a87-48fc-8801-344214a3599a",
+    "version": 1,
+    "metadata": {
+        "timestamp": "2025-06-05T09:35:20.993969+00:00",
+        "component": {
+            "type": "machine-learning-model",
+            "bom-ref": "google/owlv2-base-patch16-ensemble-99be0ddf-a3ae-5eb3-a97f-abeb17e238e5",
+            "name": "google/owlv2-base-patch16-ensemble",
+            "externalReferences": [
+                {
+                    "url": "https://huggingface.co/google/owlv2-base-patch16-ensemble",
+                    "type": "documentation"
+                }
+            ],
+            "modelCard": {
+                "modelParameters": {
+                    "task": "zero-shot-object-detection",
+                    "architectureFamily": "owlv2",
+                    "modelArchitecture": "Owlv2ForObjectDetection"
+                },
+                "properties": [
+                    {
+                        "name": "library_name",
+                        "value": "transformers"
+                    }
+                ],
+                "consideration": {
+                    "useCases": "The model is intended as a research output for research communities. We hope that this model will enable researchers to better understand and explore zero-shot, text-conditioned object detection. We also hope it can be used for interdisciplinary studies of the potential impact of such models, especially in areas that commonly require identifying objects whose label is unavailable during training."
+                }
+            },
+            "authors": [
+                {
+                    "name": "google"
+                }
+            ],
+            "licenses": [
+                {
+                    "license": {
+                        "id": "Apache-2.0",
+                        "url": "https://spdx.org/licenses/Apache-2.0.html"
+                    }
+                }
+            ],
+            "description": "The OWLv2 model (short for Open-World Localization) was proposed in [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683) by Matthias Minderer, Alexey Gritsenko, Neil Houlsby. OWLv2, like OWL-ViT, is a zero-shot text-conditioned object detection model that can be used to query an image with one or multiple text queries.The model uses CLIP as its multi-modal backbone, with a ViT-like Transformer to get visual features and a causal language model to get the text features. To use CLIP for detection, OWL-ViT removes the final token pooling layer of the vision model and attaches a lightweight classification and box head to each transformer output token. Open-vocabulary classification is enabled by replacing the fixed classification layer weights with the class-name embeddings obtained from the text model. The authors first train CLIP from scratch and fine-tune it end-to-end with the classification and box heads on standard detection datasets using a bipartite matching loss. One or multiple text queries per image can be used to perform zero-shot text-conditioned object detection.",
+            "tags": [
+                "transformers",
+                "pytorch",
+                "safetensors",
+                "owlv2",
+                "zero-shot-object-detection",
+                "vision",
+                "arxiv:2306.09683",
+                "license:apache-2.0",
+                "region:us"
+            ]
+        }
+    }
+}