There are four different versions of datasets in this repository:
MMIFR-D: This dataset is structured for detection tasks, featuring 129 subclasses distributed across five main categories.
MMIFR-FS: Tailored for few-shot learning tasks, this dataset contains 5,907 images spanning 129 distinct classes.
MMIFR-P: Designed for image-text pairing tasks, this dataset comprises 192 images, each carefully matched to a specific label.
MMIFR-OD: Intended for object detection… See the full description on the dataset page:
https://huggingface.co/datasets/HERBERT1/MMIFR.
