Do I have to resize my images?

Hello, I have database with so many images... I have to do object detection with these images... there are no annotations in this database so I think I have to use ImageLabeler to annotate the images... but when i finished annotation for example on 500 images, do I have to resize them before doing next steps ? Is this the first step before object detection? I am going to do Object detection using YOLOv4...

Answer 1

[Image Analyst]

All the original images and labeled images need to match the image size that YOLO expects.

[Adrian Kleffler]

So should I first resize all images which I am going to use and then start anottation?

[Image Analyst]

I don't know anything about databases. Do you have the database toolbox? What brand of database do you have? Oracle, Mongo, MS Access, etc.? But I don't see how some field in a database has any effect on the images you feed into your deep learning training function.

[Adrian Kleffler]

I have my own images which i was given... let´s say i have 400 images downloaded in my pc and i have to work with them ...

[Image Analyst]

Like I said. You have to resize your images to match what your neural network is expecting.

And I don't know anything about relational databases.

Answer 2

[Kevin Holly]

To answer your question in regards to databases:

https://www.mathworks.com/matlabcentral/answers/449046-how-can-i-resize-image-stored-in-imagedatastore