beetleflow-color-mask
All format examples · Machine-readable schema · Convert outputs to COCO
Directory illustration, not author JSON. Palette colors identify body parts, not individual insects. Instance extraction requires an additional algorithm.
Example
Illustrative values, not predictions from a model. Unless stated otherwise, the image is 200 pixels wide and 100 pixels high.
Readable projection: arrays and objects below illustrate the contents. The author returns Python objects/tensors or files, rather than this JSON document.
{
"predicted_mask_images": {
"0.png": "\u003coverlay image file\u003e"
},
"predicted_masks": {
"0.png": "\u003cRGB palette mask file\u003e"
}
}Convert to a standard format
This format is documented, but a conversion adapter is not yet registered.
General conversion example and required context
How to interpret it
Scope: Tested 5-class checkpoint. The hard-coded six-color palette is not a validated 9-class export.
Mask: Saved mask is palette-colored RGB, not the internal integer class map. OpenCV reads it in BGR order.
Representation: Structure describes two output directories. Each maps <folder-index>.<input-extension> filenames to raster files. Decoded RGB pixels have type T[image]: original_height rows, original_width pixels per row, three integer channels (0 through 255). These are files, not a JSON object.