These are scripts designed to take sets of images rendered from game engines (one of which represents the source image, and the other the annotations on that source image of objects of interest)-- and produce annotated datasets in various formats for ML training.
There are two versions of these scripts: one in Python, and the other in Rust. Running details for each of the scripts are in the respective folders.
Produces Supervisely annotations with both bounding box and bitmap annotations.
Currently works minimally to produce Supervisely datasets with only bounding boxes (no bitmap annotations yet).
Supervisely datasets generated with either version can then be further converted to Pascal VOC with these scripts.