What is curriculum learning in deep vision models?

Curriculum learning trains models from easy samples to harder ones gradually.

Inspired by human learning, curriculum learning presents simpler examples first and progressively introduces harder samples. In vision, this can mean starting with clean images before noisy ones or simple objects before complex scenes. It stabilizes training and improves convergence in deep architectures.