Convolutional neural networks

Knows a CNN extracts spatial patterns via learnable filters; has seen LeNet or ResNet mentioned.

Implements a basic CNN in PyTorch for image classification; understands conv, pool, stride, padding.

Fine-tunes ResNet/EfficientNet; applies transfer learning; builds detection heads on top of feature pyramids.

Designs custom architectures; understands receptive field analysis, depthwise convolutions, attention in vision (ViT).

Contributes to vision architecture research (e.g., efficient networks, multi-scale, self-supervised vision).