Image-driven object detection using OWLv2

[sokovninn]

Image-Driven Object Detection with OWLv2

This PR adds support for using images as visual prompts in OWLv2Annotator for object detection.

Example Usage:

annotator = OWLv2Annotator(device="cpu", size="base")

url = "http://images.cocodataset.org/val2017/000000039769.jpg"
im = Image.open(requests.get(url, stream=True).raw)

query_url = "http://images.cocodataset.org/val2017/000000058111.jpg"
query_image = Image.open(requests.get(query_url, stream=True).raw)

final_boxes, final_scores, final_labels = annotator.annotate_batch(
    [im], [query_image], conf_threshold=0.9
)

One query image is expected per target image to be queried. The number of query images must be either 1 (the same query image for all target images will be used) or the same as the number of target images.

In the future, integration into the main pipeline is possible.

[github-actions]

Test Results

  6 files    6 suites   2h 7m 5s ⏱️
 75 tests  57 ✅  18 💤 0 ❌
450 runs  338 ✅ 112 💤 0 ❌

Results for commit 736cfcb.

♻️ This comment has been updated with latest results.

[HonzaCuhel]

LGTM, but we need to find the cause of the failed Ubuntu unittests.

[github-actions]

☂️ Python Coverage

current status: ✅

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
1823	1125	62%	0%	🟢

New Files

No new covered files...

Modified Files

File	Coverage	Status
datadreamer/dataset_annotation/owlv2_annotator.py	65%	🟢
TOTAL	65%	🟢

updated for commit: 736cfcb by action🐍