Detections to DetectionDataset accumulator

[hardikdava]

An idea is to create a data pool accumulator which collects the results from models image, image-path, and sv.Detections

Pseudocode:

• Image Generator source i.e. source can be anything from stored images, camera stream, etc.
• Perform Inference and convert the inference output to sv.Detections
• Add image and Detection class information to Dataset pool.
• Store the Dataset using supported format yolo, coco, pascal-voc

Proposed usage

import supervision as sv

ds = sv.DetectionDataset.empty(...)

image =  cv2.imread(`image-path`)
inference_output = model(img)  # can be anything i.e. SAM, Dino, transformers, yolo, etc
detections = sv.Detection.from_model(inference_output)

image_path = "1.jpg"
ds.add_data(image_path=image_path, image=image, annotations=detections)

ds.as_coco(output_image_directory, annotation_path)

Note: This idea is not limited to only DetectionDataset but can be applied to other type of Dataset.