This feature updates the supported supervision version from 0.21.0 to 0.25.0 (latest).

Changes:

• Dataset now accepts a list of image names rather than name->image mapping. This way, images are loaded lazily, saving a massive amount of RAM. I expect issues such as the following to be fixed: numpy._core._exceptions._ArrayMemoryError: Unable to allocate 11.9 MiB for an array with shape (6, 1080, 1920) and data type bool roboflow/notebooks#336
• Instead of BoundingBoxAnnotator, using BoxAnnotator (v2)
• Instead of the extremely old BoxAnnotator (v1) that used to accept labels, now using the new BoxAnnotator + LabelAnnotator. I don't think v1 BoxAnnotator was supported in 0.21.0.

Misc changes:

• Added .vscode to .gitignore
• type annotations: np.array -> np.ndarray

Testing

Colab: https://drive.google.com/file/d/15hq88VsNfe0ViKmLcEWiF1WOa85OqT-V/view?usp=sharing