Skip to main content

Modern Architectures for Computer Vision Engineers

Modern architectures are the post-CNN designs now common in vision, chiefly attention-based models such as the Vision Transformer, which splits an image into patches and processes them with self-attention. The unit also covers how large models are trained: transfer learning, mixed precision, distributed training and tuning, which matter as much as the architecture.

Stage: Deep Vision · Part of the Computer Vision Engineer roadmap · Last reviewed 19 November 2026

What to learn in Modern Architectures

Modern Architectures interview questions

How does a Vision Transformer differ from a CNN?

A Vision Transformer splits an image into fixed-size patches, embeds them and applies self-attention so every patch can relate to every other from the first layer. A CNN builds receptive field gradually with local filters. ViTs usually need more data or strong pretraining to match CNNs.

Why does transfer learning work for images?

Early layers of a network trained on a large dataset learn general features like edges, textures and shapes that are useful across tasks. Reusing them and fine-tuning later layers lets you reach good accuracy on a small dataset with far less data and compute.

Where Modern Architectures fits in the Computer Vision Engineer roadmap

Modern Architectures belongs to the Deep Vision stage (advanced level), which takes about 38 hours. CNNs and attention.

Back to the full Computer Vision Engineer roadmap