Which of the following statements correctly describe the importance of network depth in neural networks for visual recognition tasks? Select all that apply.

Deeper neural networks can build a hierarchy of features, so early layers detect simple edges while later layers recognize complex shapes like faces or cars; this lets the model solve visual tasks that need many levels of abstraction. Because each added layer expands the space of functions the network can approximate, depth increases the expressive power without adding more parameters per se. However, very deep nets can suffer from vanishing gradients, so techniques like residual connections are used to keep learning effective. For example, a 5‑layer network might only learn edges, whereas a 20‑layer network can combine those edges into eyes, noses, and whole faces. Thus, depth is important for capturing intricate visual patterns and achieving state‑of‑the‑art performance.