A sufficient training dataset to load into ODHUB includes:
- Image files (same jpg, png, or bmp format); recommended 300 x 300-pixel image at minimum
- Label files for each image (TXT format)
- The training dataset must be placed inside a folder named "data" to meet the ODHUB format.
ODHUB also includes a built-in ANS Labelling Tool:
- If your dataset doesn't have label files, you can use the Labelling Tool to label your images
- If your dataset already has label files in different formats (XML, JSON), the Labelling Tool supports generating additional TXT label files from the original format so you can quickly reuse your label information.