[0.3.0] - 2023-03-06

[SkalskiP]

🚀 Added

New methods in sv.Detections API:

• from_transformers - convert Object Detection 🤗 Transformer result into sv.Detections
• from_detectron2 - convert Detectron2 result into sv.Detections
• from_coco_annotations - convert COCO annotation into sv.Detections
• area - dynamically calculated property storing bbox area
• with_nms - initial implementation (only class agnostic) of sv.Detections NMS

🌱 Changed

• Make sv.Detections.confidence field Optional.

🏆 Contributors

• @SkalskiP

---

This discussion was created from the release [0.3.0] - 2023-03-06.

[mohamedaliELfeky]

In the from_coco_annotations

I get this error:

AttributeError: type object 'Detections' has no attribute 'from_coco_annotations'

[onuralpszr]

Hello @mohamedaliELfeky first off, If you are using 0.3.0 version please update to latest first. Secondly "from_coco_annotations" has been renamed and move into dataset section and for using loading coco

https://github.com/roboflow/supervision/blob/develop/supervision/dataset/core.py

           >>> import roboflow
           >>> from roboflow import Roboflow
           >>> import supervision as sv

           >>> roboflow.login()

           >>> rf = Roboflow()

           >>> project = rf.workspace(WORKSPACE_ID).project(PROJECT_ID)
           >>> dataset = project.version(PROJECT_VERSION).download("coco")

           >>> ds = sv.DetectionDataset.from_coco(
           ...     images_directory_path=f"{dataset.location}/train",
           ...     annotations_path=f"{dataset.location}/train/_annotations.coco.json",
           ... )

           >>> ds.classes
           ['dog', 'person']

[onuralpszr]

We should be also extra careful for release notes

https://github.com/roboflow/supervision/releases/tag/0.11.0 no deprecation or removal warning message in "release notes"

Sorry for confiusion

[Gitgutgait]

Hello @mohamedaliELfeky first off, If you are using 0.3.0 version please update to latest first. Secondly "from_coco_annotations" has been renamed and move into dataset section and for using loading coco

https://github.com/roboflow/supervision/blob/develop/supervision/dataset/core.py

           >>> import roboflow
           >>> from roboflow import Roboflow
           >>> import supervision as sv

           >>> roboflow.login()

           >>> rf = Roboflow()

           >>> project = rf.workspace(WORKSPACE_ID).project(PROJECT_ID)
           >>> dataset = project.version(PROJECT_VERSION).download("coco")

           >>> ds = sv.DetectionDataset.from_coco(
           ...     images_directory_path=f"{dataset.location}/train",
           ...     annotations_path=f"{dataset.location}/train/_annotations.coco.json",
           ... )

           >>> ds.classes
           ['dog', 'person']

But if I change it like this :

detections = sv.DetectionDataset.from_coco(images_directory_path=TRAIN_DIRECTORY,
annotations_path=f"{TRAIN_DIRECTORY}/annotations.json")

Then i will get error for the labels:

we will use id2label function for training

categories = TRAIN_DATASET.coco.cats
id2label = {k: v['name'] for k,v in categories.items()}
labels = [
f"{id2label[class_id]}"
for _, _, class_id, _
in detections
]

ValueError: not enough values to unpack (expected 4, got 3)

[hebrotem]

this might not be an efficient way, but it helped me:

[Gitgutgait]

yes, it plots all the images in ds, but the original purpose was to plot the random selected image.

When using from_coco like this :
detections = sv.DetectionDataset.from_coco(images_directory_path = TRAIN_DIRECTORY,
annotations_path=f"{TRAIN_DIRECTORY}/annotations.json")
it take all the images and annotations from the directory

@SkalskiP, btw it didn't work, now it says :
ValueError: not enough values to unpack (expected 5, got 3)

[Gitgutgait]

If the last 4 statements is outside of the loop then it works as it should!

[hebrotem]

I doubt if trying that would make it work as it should.

[Gitgutgait]

you were right, it didn't work as it should, but this is the updated code:

find the image path you want in the detections, and then use that one :D

[alandevkota]

I solved similar errors with this code: