Create new training variations of each image so the model sees more variety: small rotations, horizontal flips, small zoom and shifts. Only on training data, never on validation/test. Do not use vertical flips (an upside-down face is not normal).
Notes, key code, what to skip and 1 curated resources are inside the platform.Join the waitlist and we'll build you a roadmap that includes this topic only if your goal needs it.