Images Dataset Conversion into h5 format in Pytorch

I require the conversion of an image dataset from its current format (.png) to some sort of memory mapped file like HDF5 (aka H5) or LMDB and embedding in my code successfully
My code in Python language using (Pytorch) written in Google Colab

- The dataset will be shared upon project commencement.
- The purpose of the conversion is to optimize the deep learning model training in speed
- The important point in the code I have used the k-fold cross validation, so I want that worked with the new dataset format in professional way
- Run the code as work well with new dataset format successfully

Ideal Freelancer:
- Proficient in image processing and conversion techniques.
- Familiarity with data analysis and deep learning applications.

Thanks