Currently, I am working on the label printing defect detection project. My project is mainly focus on finding the location of defect on the label printing. Can I know whether there is any method/ way to change the brightness of the validation and testing dataset? Because I want to identify (through validation and testing) whether the YOLO model that I trained can show good result on different brightness or not.
For example, I want to change the brightness of validation and testing dataset to ± 10%. Is this possible? Thank You to who which is willing to help me.
Hello! You can see more information about augmentations here: Generate Image Augmentations with Roboflow
One way to alter the brightness of your dataset would be to apply only the brightness augmentation and generate a dataset. Then export that data and upload to validation and test datasets.
Okay! Right now I get some idea on how to do it. Thank You So Much.
Actually, now I am curious, during brightness augmentation (25%) and x2 dataset, can I ensure original of ‘Image 1’, after augmentation the ‘image 1’ will become +25% brightness and -25% brightness?
Because during I review the dataset after brightness augmentation, some of the image will not change it brightness.