There are many features in an image dataset which a can correctly classify the dataset. Neural nets usually focus in texture which we human clearly don't, we focus on a higher level abstraction and usually ignore texture. Using either would encourage the network to use broader features and ignore texture
Why is this not a typical data augmentation method for training deep networks? It would allow the networks to learn invariants that's align with humans. Currently, I suppose none of these dithered images would be correctly classified.
The authors experience is not mine. I have an app/website combo deployed that is getting 60+ visitors a day and the app is getting downloaded weekly. Both are growing. All "vibe" coded.
Now I have 20 YOE in software too before the booon of AI.
I built a synthetic aperture sonar / radar simulation engine so I could generate endless amounts of procedural generated scenes to train a world model. Data for sonar is especially hard to come by.
Too many codes or old or gate kept behind proprietary walls. Many are old and don't use the newest acceleration techniquea to make the simulation fast. Additionally, none of them scale using aws. I want SAS/SAR image to be easy to generate for anyone.
I have a similar finding for a website I made that collates college town bar specials and live music. Using agents with vision models works but it's not as straightforward as one would initially think. U can check out the results here. https://www.nittanynights.com