Importing CelebA dataset (deep learning) into MATLAB for use with YOLO

Hi,

Following the example on the MATLAB website for object detection using YOLO V3, I'd like to replace the dataset provided with the example (vehicle Dataset) with a faces dataset called CelebA.

Here's a link to the dataset on kaggle:

https://www.kaggle.com/jessicali9530/celeba-dataset

And a link to the Yolo V3 example:

Object Detection Using YOLO v3 Deep Learning - MATLAB & Simulink (mathworks.com)

I've downloaded the images of the dataset as well as a list_bbox_celeba.csv file, which, as its name suggests, contains bounding box data for the images.

How can I import this dataset (images + bounding box data) into MATLAB so that I could use the Yolo V3 example code for face detection instead of vehicle detection?

Thanks.

Answer 1

[Mahesh Taparia]

Hi

The training data can have 2 formats either tabular or in the form of datastore as given here. In your case, create a table with file/ image path in 1st column and remaining columns with box labels as indicated in the above link. So somedata preprocessing is required before passing to an object detector.

Hope it will help!